Download manifests/merge_step_00002000.json from mrjackspade/Ideogram4-Natural-Language-Text-Encoder: direct link, hf CLI and curl.
- Browser
- Download file 115 kB
-
https://huggingface.co/mrjackspade/Ideogram4-Natural-Language-Text-Encoder/resolve/main/manifests/merge_step_00002000.json
- Command line
-
hf download hf://mrjackspade/Ideogram4-Natural-Language-Text-Encoder/manifests/merge_step_00002000.json
-
curl -L -o merge_step_00002000.json https://huggingface.co/mrjackspade/Ideogram4-Natural-Language-Text-Encoder/resolve/main/manifests/merge_step_00002000.json
115 kB
| { | |
| "schema_version": 1, | |
| "format": "comfy_scaled_fp8_merged_text_encoder", | |
| "scale_policy": "preserve_stock_per_tensor_scale", | |
| "base": "/workspace/Ideogram/checkpoints/text_encoders/qwen3vl_8b_fp8_scaled.safetensors", | |
| "base_sha256": "4ba424cf62e51392e4d1a39933e803706f4e823c1065f36aaf149c6453f66bcd", | |
| "adapter": "/workspace/Ideogram/artifacts/phase2_balanced_4000_8state_debanner_s020_v1_checkpoints/step_00002000/adapters.safetensors", | |
| "adapter_sha256": "ade88e4b6928ca2556a868b566e7bccd3d927e80a0d3847ebd4583fab10bc63d", | |
| "adapter_global_step": "2000", | |
| "output": "/workspace/Ideogram/artifacts/huggingface_release/s020_v1/qwen3vl_8b_ideogram4_nl_s020_v1_step_00002000_fp8_scaled.safetensors", | |
| "output_bytes": 10588637800, | |
| "output_sha256": "5c505aec1b9aacac4483af9b09a0d3cd2d62a54b15ae3afbe83cf7945012eb98", | |
| "output_tensors": 1254, | |
| "merged_projections": 252, | |
| "lora_rank": 64, | |
| "lora_alpha": 64.0, | |
| "elapsed_seconds": 61.729, | |
| "projections": [ | |
| { | |
| "index": 1, | |
| "total": 252, | |
| "projection": "layers.0.mlp.down_proj", | |
| "shape": [ | |
| 4096, | |
| 12288 | |
| ], | |
| "seconds": 0.4391, | |
| "parameters": 50331648, | |
| "reference_delta_l2": 3.1295157799351236, | |
| "retained_delta_l2": 2.8502838015288536, | |
| "delta_retention_norm_ratio": 0.9107747018894857, | |
| "delta_cosine": 0.7261043086648739, | |
| "relative_weight_rmse": 0.012313553632253754 | |
| }, | |
| { | |
| "index": 2, | |
| "total": 252, | |
| "projection": "layers.0.mlp.gate_proj", | |
| "shape": [ | |
| 12288, | |
| 4096 | |
| ], | |
| "seconds": 0.3781, | |
| "parameters": 50331648, | |
| "reference_delta_l2": 3.308858942324612, | |
| "retained_delta_l2": 3.0219341535998163, | |
| "delta_retention_norm_ratio": 0.9132858808048134, | |
| "delta_cosine": 0.7258020356483821, | |
| "relative_weight_rmse": 0.013496873111315233 | |
| }, | |
| { | |
| "index": 3, | |
| "total": 252, | |
| "projection": "layers.0.mlp.up_proj", | |
| "shape": [ | |
| 12288, | |
| 4096 | |
| ], | |
| "seconds": 0.38, | |
| "parameters": 50331648, | |
| "reference_delta_l2": 2.9333920986669484, | |
| "retained_delta_l2": 2.621853415304213, | |
| "delta_retention_norm_ratio": 0.893795758328963, | |
| "delta_cosine": 0.7068022139752237, | |
| "relative_weight_rmse": 0.01302276820099369 | |
| }, | |
| { | |
| "index": 4, | |
| "total": 252, | |
| "projection": "layers.0.self_attn.k_proj", | |
| "shape": [ | |
| 1024, | |
| 4096 | |
| ], | |
| "seconds": 0.0304, | |
| "parameters": 4194304, | |
| "reference_delta_l2": 0.602349856468238, | |
| "retained_delta_l2": 0.44131584785155675, | |
| "delta_retention_norm_ratio": 0.7326570150429302, | |
| "delta_cosine": 0.560430668071971, | |
| "relative_weight_rmse": 0.009007892007776262 | |
| }, | |
| { | |
| "index": 5, | |
| "total": 252, | |
| "projection": "layers.0.self_attn.o_proj", | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "seconds": 0.1278, | |
| "parameters": 16777216, | |
| "reference_delta_l2": 1.9730080764510292, | |
| "retained_delta_l2": 1.8610741013841146, | |
| "delta_retention_norm_ratio": 0.9432673508015957, | |
| "delta_cosine": 0.762789735772714, | |
| "relative_weight_rmse": 0.013622010741295177 | |
| }, | |
| { | |
| "index": 6, | |
| "total": 252, | |
| "projection": "layers.0.self_attn.q_proj", | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "seconds": 0.127, | |
| "parameters": 16777216, | |
| "reference_delta_l2": 1.223923628071747, | |
| "retained_delta_l2": 0.9518550472806543, | |
| "delta_retention_norm_ratio": 0.7777078777213182, | |
| "delta_cosine": 0.598442627196136, | |
| "relative_weight_rmse": 0.009797371165960555 | |
| }, | |
| { | |
| "index": 7, | |
| "total": 252, | |
| "projection": "layers.0.self_attn.v_proj", | |
| "shape": [ | |
| 1024, | |
| 4096 | |
| ], | |
| "seconds": 0.0291, | |
| "parameters": 4194304, | |
| "reference_delta_l2": 0.7957482199785566, | |
| "retained_delta_l2": 0.6933537877647481, | |
| "delta_retention_norm_ratio": 0.8713230772711401, | |
| "delta_cosine": 0.6841275851947753, | |
| "relative_weight_rmse": 0.011987200008821542 | |
| }, | |
| { | |
| "index": 8, | |
| "total": 252, | |
| "projection": "layers.1.mlp.down_proj", | |
| "shape": [ | |
| 4096, | |
| 12288 | |
| ], | |
| "seconds": 0.4182, | |
| "parameters": 50331648, | |
| "reference_delta_l2": 2.0362647955657387, | |
| "retained_delta_l2": 1.9007486242357023, | |
| "delta_retention_norm_ratio": 0.933448649888196, | |
| "delta_cosine": 0.7565510162524389, | |
| "relative_weight_rmse": 0.011982820281728953 | |
| }, | |
| { | |
| "index": 9, | |
| "total": 252, | |
| "projection": "layers.1.mlp.gate_proj", | |
| "shape": [ | |
| 12288, | |
| 4096 | |
| ], | |
| "seconds": 0.3785, | |
| "parameters": 50331648, | |
| "reference_delta_l2": 3.8319098718741618, | |
| "retained_delta_l2": 3.798833927371201, | |
| "delta_retention_norm_ratio": 0.9913682874574543, | |
| "delta_cosine": 0.8459146309675251, | |
| "relative_weight_rmse": 0.014360352713851662 | |
| }, | |
| { | |
| "index": 10, | |
| "total": 252, | |
| "projection": "layers.1.mlp.up_proj", | |
| "shape": [ | |
| 12288, | |
| 4096 | |
| ], | |
| "seconds": 0.3812, | |
| "parameters": 50331648, | |
| "reference_delta_l2": 4.233617448400264, | |
| "retained_delta_l2": 4.23070853812047, | |
| "delta_retention_norm_ratio": 0.9993129019531762, | |
| "delta_cosine": 0.929609242827706, | |
| "relative_weight_rmse": 0.015762857213329554 | |
| }, | |
| { | |
| "index": 11, | |
| "total": 252, | |
| "projection": "layers.1.self_attn.k_proj", | |
| "shape": [ | |
| 1024, | |
| 4096 | |
| ], | |
| "seconds": 0.0305, | |
| "parameters": 4194304, | |
| "reference_delta_l2": 0.6761311740026398, | |
| "retained_delta_l2": 0.5551743744851045, | |
| "delta_retention_norm_ratio": 0.8211045368586083, | |
| "delta_cosine": 0.6373080054133655, | |
| "relative_weight_rmse": 0.010814861671689595 | |
| }, | |
| { | |
| "index": 12, | |
| "total": 252, | |
| "projection": "layers.1.self_attn.o_proj", | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "seconds": 0.1282, | |
| "parameters": 16777216, | |
| "reference_delta_l2": 1.7147328050292834, | |
| "retained_delta_l2": 1.5341219402892992, | |
| "delta_retention_norm_ratio": 0.8946711323127106, | |
| "delta_cosine": 0.7053989373758582, | |
| "relative_weight_rmse": 0.012667201785381724 | |
| }, | |
| { | |
| "index": 13, | |
| "total": 252, | |
| "projection": "layers.1.self_attn.q_proj", | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "seconds": 0.1276, | |
| "parameters": 16777216, | |
| "reference_delta_l2": 2.0926542666331747, | |
| "retained_delta_l2": 2.023345819293772, | |
| "delta_retention_norm_ratio": 0.9668801251862249, | |
| "delta_cosine": 0.7927101763251782, | |
| "relative_weight_rmse": 0.014636375532536479 | |
| }, | |
| { | |
| "index": 14, | |
| "total": 252, | |
| "projection": "layers.1.self_attn.v_proj", | |
| "shape": [ | |
| 1024, | |
| 4096 | |
| ], | |
| "seconds": 0.0293, | |
| "parameters": 4194304, | |
| "reference_delta_l2": 0.9182738031034758, | |
| "retained_delta_l2": 0.8358768448352694, | |
| "delta_retention_norm_ratio": 0.910269727841815, | |
| "delta_cosine": 0.7230853742697265, | |
| "relative_weight_rmse": 0.012467732033120576 | |
| }, | |
| { | |
| "index": 15, | |
| "total": 252, | |
| "projection": "layers.10.mlp.down_proj", | |
| "shape": [ | |
| 4096, | |
| 12288 | |
| ], | |
| "seconds": 0.4181, | |
| "parameters": 50331648, | |
| "reference_delta_l2": 2.8113782214961516, | |
| "retained_delta_l2": 2.492795325814636, | |
| "delta_retention_norm_ratio": 0.8866808836870148, | |
| "delta_cosine": 0.7006694785539181, | |
| "relative_weight_rmse": 0.011660723877415962 | |
| }, | |
| { | |
| "index": 16, | |
| "total": 252, | |
| "projection": "layers.10.mlp.gate_proj", | |
| "shape": [ | |
| 12288, | |
| 4096 | |
| ], | |
| "seconds": 0.3776, | |
| "parameters": 50331648, | |
| "reference_delta_l2": 2.388968981485157, | |
| "retained_delta_l2": 1.8515130111340723, | |
| "delta_retention_norm_ratio": 0.7750259737499969, | |
| "delta_cosine": 0.5958672482894636, | |
| "relative_weight_rmse": 0.009648177171664604 | |
| }, | |
| { | |
| "index": 17, | |
| "total": 252, | |
| "projection": "layers.10.mlp.up_proj", | |
| "shape": [ | |
| 12288, | |
| 4096 | |
| ], | |
| "seconds": 0.3793, | |
| "parameters": 50331648, | |
| "reference_delta_l2": 2.8546671366650447, | |
| "retained_delta_l2": 2.525883959859384, | |
| "delta_retention_norm_ratio": 0.8848260896751133, | |
| "delta_cosine": 0.6973434819774882, | |
| "relative_weight_rmse": 0.011903683508779728 | |
| }, | |
| { | |
| "index": 18, | |
| "total": 252, | |
| "projection": "layers.10.self_attn.k_proj", | |
| "shape": [ | |
| 1024, | |
| 4096 | |
| ], | |
| "seconds": 0.031, | |
| "parameters": 4194304, | |
| "reference_delta_l2": 0.898296856891836, | |
| "retained_delta_l2": 0.8255322300406948, | |
| "delta_retention_norm_ratio": 0.9189971262919572, | |
| "delta_cosine": 0.7372096005027154, | |
| "relative_weight_rmse": 0.011972097540133963 | |
| }, | |
| { | |
| "index": 19, | |
| "total": 252, | |
| "projection": "layers.10.self_attn.o_proj", | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "seconds": 0.1294, | |
| "parameters": 16777216, | |
| "reference_delta_l2": 1.773726814891075, | |
| "retained_delta_l2": 1.5849685945727472, | |
| "delta_retention_norm_ratio": 0.8935810076649715, | |
| "delta_cosine": 0.7068052779610151, | |
| "relative_weight_rmse": 0.012165208402044565 | |
| }, | |
| { | |
| "index": 20, | |
| "total": 252, | |
| "projection": "layers.10.self_attn.q_proj", | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "seconds": 0.1269, | |
| "parameters": 16777216, | |
| "reference_delta_l2": 2.1181889970136627, | |
| "retained_delta_l2": 2.020497366449768, | |
| "delta_retention_norm_ratio": 0.9538796440253322, | |
| "delta_cosine": 0.7798677582043131, | |
| "relative_weight_rmse": 0.013449250140918962 | |
| }, | |
| { | |
| "index": 21, | |
| "total": 252, | |
| "projection": "layers.10.self_attn.v_proj", | |
| "shape": [ | |
| 1024, | |
| 4096 | |
| ], | |
| "seconds": 0.0289, | |
| "parameters": 4194304, | |
| "reference_delta_l2": 0.9094617335032259, | |
| "retained_delta_l2": 0.8120646973699481, | |
| "delta_retention_norm_ratio": 0.8929069442448044, | |
| "delta_cosine": 0.7111358371007716, | |
| "relative_weight_rmse": 0.011201089063293127 | |
| }, | |
| { | |
| "index": 22, | |
| "total": 252, | |
| "projection": "layers.11.mlp.down_proj", | |
| "shape": [ | |
| 4096, | |
| 12288 | |
| ], | |
| "seconds": 0.4199, | |
| "parameters": 50331648, | |
| "reference_delta_l2": 2.6083400962198855, | |
| "retained_delta_l2": 2.2326133808508537, | |
| "delta_retention_norm_ratio": 0.8559517925160334, | |
| "delta_cosine": 0.6705604386194571, | |
| "relative_weight_rmse": 0.011045630389532286 | |
| }, | |
| { | |
| "index": 23, | |
| "total": 252, | |
| "projection": "layers.11.mlp.gate_proj", | |
| "shape": [ | |
| 12288, | |
| 4096 | |
| ], | |
| "seconds": 0.384, | |
| "parameters": 50331648, | |
| "reference_delta_l2": 2.7719414238376725, | |
| "retained_delta_l2": 2.339405933084685, | |
| "delta_retention_norm_ratio": 0.8439593683209384, | |
| "delta_cosine": 0.659378819125765, | |
| "relative_weight_rmse": 0.010807064801439835 | |
| }, | |
| { | |
| "index": 24, | |
| "total": 252, | |
| "projection": "layers.11.mlp.up_proj", | |
| "shape": [ | |
| 12288, | |
| 4096 | |
| ], | |
| "seconds": 0.3814, | |
| "parameters": 50331648, | |
| "reference_delta_l2": 2.6728273375827234, | |
| "retained_delta_l2": 2.2706054481938365, | |
| "delta_retention_norm_ratio": 0.8495144509586422, | |
| "delta_cosine": 0.6631438140090375, | |
| "relative_weight_rmse": 0.011368344170365544 | |
| }, | |
| { | |
| "index": 25, | |
| "total": 252, | |
| "projection": "layers.11.self_attn.k_proj", | |
| "shape": [ | |
| 1024, | |
| 4096 | |
| ], | |
| "seconds": 0.0303, | |
| "parameters": 4194304, | |
| "reference_delta_l2": 1.1285426297564844, | |
| "retained_delta_l2": 1.0856459717547504, | |
| "delta_retention_norm_ratio": 0.9619893330826232, | |
| "delta_cosine": 0.7870521772506621, | |
| "relative_weight_rmse": 0.01365816659419093 | |
| }, | |
| { | |
| "index": 26, | |
| "total": 252, | |
| "projection": "layers.11.self_attn.o_proj", | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "seconds": 0.128, | |
| "parameters": 16777216, | |
| "reference_delta_l2": 1.6698245744642877, | |
| "retained_delta_l2": 1.466318444183154, | |
| "delta_retention_norm_ratio": 0.8781272395955592, | |
| "delta_cosine": 0.693043112285495, | |
| "relative_weight_rmse": 0.011724965342670564 | |
| }, | |
| { | |
| "index": 27, | |
| "total": 252, | |
| "projection": "layers.11.self_attn.q_proj", | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "seconds": 0.1296, | |
| "parameters": 16777216, | |
| "reference_delta_l2": 1.4064437305620288, | |
| "retained_delta_l2": 1.169695210234028, | |
| "delta_retention_norm_ratio": 0.8316686866431593, | |
| "delta_cosine": 0.6464514802859761, | |
| "relative_weight_rmse": 0.010750990640851749 | |
| }, | |
| { | |
| "index": 28, | |
| "total": 252, | |
| "projection": "layers.11.self_attn.v_proj", | |
| "shape": [ | |
| 1024, | |
| 4096 | |
| ], | |
| "seconds": 0.0292, | |
| "parameters": 4194304, | |
| "reference_delta_l2": 0.9976261903719174, | |
| "retained_delta_l2": 0.9075747276461306, | |
| "delta_retention_norm_ratio": 0.9097342636000612, | |
| "delta_cosine": 0.7270924978235055, | |
| "relative_weight_rmse": 0.011884848481938047 | |
| }, | |
| { | |
| "index": 29, | |
| "total": 252, | |
| "projection": "layers.12.mlp.down_proj", | |
| "shape": [ | |
| 4096, | |
| 12288 | |
| ], | |
| "seconds": 0.4185, | |
| "parameters": 50331648, | |
| "reference_delta_l2": 2.5673767174764017, | |
| "retained_delta_l2": 2.1877978222092476, | |
| "delta_retention_norm_ratio": 0.8521530195848076, | |
| "delta_cosine": 0.6667567491113927, | |
| "relative_weight_rmse": 0.010833193858063861 | |
| }, | |
| { | |
| "index": 30, | |
| "total": 252, | |
| "projection": "layers.12.mlp.gate_proj", | |
| "shape": [ | |
| 12288, | |
| 4096 | |
| ], | |
| "seconds": 0.3797, | |
| "parameters": 50331648, | |
| "reference_delta_l2": 3.2128146573103873, | |
| "retained_delta_l2": 2.880827242553919, | |
| "delta_retention_norm_ratio": 0.8966677352516836, | |
| "delta_cosine": 0.7127643911365393, | |
| "relative_weight_rmse": 0.011978474399065964 | |
| }, | |
| { | |
| "index": 31, | |
| "total": 252, | |
| "projection": "layers.12.mlp.up_proj", | |
| "shape": [ | |
| 12288, | |
| 4096 | |
| ], | |
| "seconds": 0.3784, | |
| "parameters": 50331648, | |
| "reference_delta_l2": 2.685915319766742, | |
| "retained_delta_l2": 2.2893747931314667, | |
| "delta_retention_norm_ratio": 0.8523629826610795, | |
| "delta_cosine": 0.6661181577537216, | |
| "relative_weight_rmse": 0.011217255943459098 | |
| }, | |
| { | |
| "index": 32, | |
| "total": 252, | |
| "projection": "layers.12.self_attn.k_proj", | |
| "shape": [ | |
| 1024, | |
| 4096 | |
| ], | |
| "seconds": 0.0302, | |
| "parameters": 4194304, | |
| "reference_delta_l2": 0.9323220288102295, | |
| "retained_delta_l2": 0.8656498962907475, | |
| "delta_retention_norm_ratio": 0.928488086241441, | |
| "delta_cosine": 0.7477878360800111, | |
| "relative_weight_rmse": 0.012288728180501179 | |
| }, | |
| { | |
| "index": 33, | |
| "total": 252, | |
| "projection": "layers.12.self_attn.o_proj", | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "seconds": 0.1282, | |
| "parameters": 16777216, | |
| "reference_delta_l2": 1.7732730109339851, | |
| "retained_delta_l2": 1.6063538404039202, | |
| "delta_retention_norm_ratio": 0.9058694462156459, | |
| "delta_cosine": 0.7218775372986568, | |
| "relative_weight_rmse": 0.012030348575707556 | |
| }, | |
| { | |
| "index": 34, | |
| "total": 252, | |
| "projection": "layers.12.self_attn.q_proj", | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "seconds": 0.1279, | |
| "parameters": 16777216, | |
| "reference_delta_l2": 1.8059135630192593, | |
| "retained_delta_l2": 1.6509890099638174, | |
| "delta_retention_norm_ratio": 0.914212642161883, | |
| "delta_cosine": 0.7311857138038845, | |
| "relative_weight_rmse": 0.012431793623686902 | |
| }, | |
| { | |
| "index": 35, | |
| "total": 252, | |
| "projection": "layers.12.self_attn.v_proj", | |
| "shape": [ | |
| 1024, | |
| 4096 | |
| ], | |
| "seconds": 0.0294, | |
| "parameters": 4194304, | |
| "reference_delta_l2": 1.0870296328963427, | |
| "retained_delta_l2": 1.0239311844644483, | |
| "delta_retention_norm_ratio": 0.9419533317930153, | |
| "delta_cosine": 0.7659336051191952, | |
| "relative_weight_rmse": 0.012192820958434733 | |
| }, | |
| { | |
| "index": 36, | |
| "total": 252, | |
| "projection": "layers.13.mlp.down_proj", | |
| "shape": [ | |
| 4096, | |
| 12288 | |
| ], | |
| "seconds": 0.4212, | |
| "parameters": 50331648, | |
| "reference_delta_l2": 2.344962195460705, | |
| "retained_delta_l2": 1.9205837316330012, | |
| "delta_retention_norm_ratio": 0.8190254560823196, | |
| "delta_cosine": 0.6358079001534217, | |
| "relative_weight_rmse": 0.010191678392086631 | |
| }, | |
| { | |
| "index": 37, | |
| "total": 252, | |
| "projection": "layers.13.mlp.gate_proj", | |
| "shape": [ | |
| 12288, | |
| 4096 | |
| ], | |
| "seconds": 0.377, | |
| "parameters": 50331648, | |
| "reference_delta_l2": 2.9876379632012293, | |
| "retained_delta_l2": 2.6129843228546905, | |
| "delta_retention_norm_ratio": 0.8745987147836679, | |
| "delta_cosine": 0.6875034086966243, | |
| "relative_weight_rmse": 0.011880131031540675 | |
| }, | |
| { | |
| "index": 38, | |
| "total": 252, | |
| "projection": "layers.13.mlp.up_proj", | |
| "shape": [ | |
| 12288, | |
| 4096 | |
| ], | |
| "seconds": 0.3865, | |
| "parameters": 50331648, | |
| "reference_delta_l2": 2.4021005717028983, | |
| "retained_delta_l2": 1.9422196707182822, | |
| "delta_retention_norm_ratio": 0.808550521821575, | |
| "delta_cosine": 0.6251684158384617, | |
| "relative_weight_rmse": 0.010350500532359528 | |
| }, | |
| { | |
| "index": 39, | |
| "total": 252, | |
| "projection": "layers.13.self_attn.k_proj", | |
| "shape": [ | |
| 1024, | |
| 4096 | |
| ], | |
| "seconds": 0.0305, | |
| "parameters": 4194304, | |
| "reference_delta_l2": 1.3411620116922038, | |
| "retained_delta_l2": 1.3268981283278296, | |
| "delta_retention_norm_ratio": 0.9893645337103033, | |
| "delta_cosine": 0.8345317232908486, | |
| "relative_weight_rmse": 0.014773482348064967 | |
| }, | |
| { | |
| "index": 40, | |
| "total": 252, | |
| "projection": "layers.13.self_attn.o_proj", | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "seconds": 0.132, | |
| "parameters": 16777216, | |
| "reference_delta_l2": 1.7296016855603318, | |
| "retained_delta_l2": 1.5499819642029007, | |
| "delta_retention_norm_ratio": 0.8961496610132866, | |
| "delta_cosine": 0.7118572199998348, | |
| "relative_weight_rmse": 0.012260288668890474 | |
| }, | |
| { | |
| "index": 41, | |
| "total": 252, | |
| "projection": "layers.13.self_attn.q_proj", | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "seconds": 0.1281, | |
| "parameters": 16777216, | |
| "reference_delta_l2": 1.4093637562418553, | |
| "retained_delta_l2": 1.1732090915470472, | |
| "delta_retention_norm_ratio": 0.8324388124436184, | |
| "delta_cosine": 0.6470332479577902, | |
| "relative_weight_rmse": 0.010748170228869565 | |
| }, | |
| { | |
| "index": 42, | |
| "total": 252, | |
| "projection": "layers.13.self_attn.v_proj", | |
| "shape": [ | |
| 1024, | |
| 4096 | |
| ], | |
| "seconds": 0.0292, | |
| "parameters": 4194304, | |
| "reference_delta_l2": 1.026313557917772, | |
| "retained_delta_l2": 0.9447760209279176, | |
| "delta_retention_norm_ratio": 0.9205529963422863, | |
| "delta_cosine": 0.7379792429350502, | |
| "relative_weight_rmse": 0.01230172862322642 | |
| }, | |
| { | |
| "index": 43, | |
| "total": 252, | |
| "projection": "layers.14.mlp.down_proj", | |
| "shape": [ | |
| 4096, | |
| 12288 | |
| ], | |
| "seconds": 0.4209, | |
| "parameters": 50331648, | |
| "reference_delta_l2": 2.467721314345068, | |
| "retained_delta_l2": 2.086833464404717, | |
| "delta_retention_norm_ratio": 0.8456519997917843, | |
| "delta_cosine": 0.661331921101294, | |
| "relative_weight_rmse": 0.01051814884350571 | |
| }, | |
| { | |
| "index": 44, | |
| "total": 252, | |
| "projection": "layers.14.mlp.gate_proj", | |
| "shape": [ | |
| 12288, | |
| 4096 | |
| ], | |
| "seconds": 0.3765, | |
| "parameters": 50331648, | |
| "reference_delta_l2": 2.7321969937263284, | |
| "retained_delta_l2": 2.3171825631863587, | |
| "delta_retention_norm_ratio": 0.8481023032040055, | |
| "delta_cosine": 0.6614208547580835, | |
| "relative_weight_rmse": 0.011356928376026088 | |
| }, | |
| { | |
| "index": 45, | |
| "total": 252, | |
| "projection": "layers.14.mlp.up_proj", | |
| "shape": [ | |
| 12288, | |
| 4096 | |
| ], | |
| "seconds": 0.3807, | |
| "parameters": 50331648, | |
| "reference_delta_l2": 2.9443813100266794, | |
| "retained_delta_l2": 2.610028726180295, | |
| "delta_retention_norm_ratio": 0.8864438574216615, | |
| "delta_cosine": 0.7008685493193013, | |
| "relative_weight_rmse": 0.011782602760381216 | |
| }, | |
| { | |
| "index": 46, | |
| "total": 252, | |
| "projection": "layers.14.self_attn.k_proj", | |
| "shape": [ | |
| 1024, | |
| 4096 | |
| ], | |
| "seconds": 0.0303, | |
| "parameters": 4194304, | |
| "reference_delta_l2": 0.7284994577708277, | |
| "retained_delta_l2": 0.618501899336792, | |
| "delta_retention_norm_ratio": 0.8490080435054507, | |
| "delta_cosine": 0.6659961735692724, | |
| "relative_weight_rmse": 0.010714954869028445 | |
| }, | |
| { | |
| "index": 47, | |
| "total": 252, | |
| "projection": "layers.14.self_attn.o_proj", | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "seconds": 0.1282, | |
| "parameters": 16777216, | |
| "reference_delta_l2": 1.5558381795834701, | |
| "retained_delta_l2": 1.354539985578835, | |
| "delta_retention_norm_ratio": 0.8706175252374081, | |
| "delta_cosine": 0.6849474569724862, | |
| "relative_weight_rmse": 0.011441734643697268 | |
| }, | |
| { | |
| "index": 48, | |
| "total": 252, | |
| "projection": "layers.14.self_attn.q_proj", | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "seconds": 0.1269, | |
| "parameters": 16777216, | |
| "reference_delta_l2": 2.318720626493867, | |
| "retained_delta_l2": 2.2602193255120104, | |
| "delta_retention_norm_ratio": 0.9747700088085574, | |
| "delta_cosine": 0.8055496838820395, | |
| "relative_weight_rmse": 0.01391485494866503 | |
| }, | |
| { | |
| "index": 49, | |
| "total": 252, | |
| "projection": "layers.14.self_attn.v_proj", | |
| "shape": [ | |
| 1024, | |
| 4096 | |
| ], | |
| "seconds": 0.029, | |
| "parameters": 4194304, | |
| "reference_delta_l2": 1.0174185656146237, | |
| "retained_delta_l2": 0.944473320821553, | |
| "delta_retention_norm_ratio": 0.9283036035920925, | |
| "delta_cosine": 0.7457334422346245, | |
| "relative_weight_rmse": 0.012384185749512564 | |
| }, | |
| { | |
| "index": 50, | |
| "total": 252, | |
| "projection": "layers.15.mlp.down_proj", | |
| "shape": [ | |
| 4096, | |
| 12288 | |
| ], | |
| "seconds": 0.4199, | |
| "parameters": 50331648, | |
| "reference_delta_l2": 2.371493147271087, | |
| "retained_delta_l2": 1.9731258260588727, | |
| "delta_retention_norm_ratio": 0.8320183544824401, | |
| "delta_cosine": 0.6481631721173573, | |
| "relative_weight_rmse": 0.010300907614621341 | |
| }, | |
| { | |
| "index": 51, | |
| "total": 252, | |
| "projection": "layers.15.mlp.gate_proj", | |
| "shape": [ | |
| 12288, | |
| 4096 | |
| ], | |
| "seconds": 0.3794, | |
| "parameters": 50331648, | |
| "reference_delta_l2": 2.8697631793277374, | |
| "retained_delta_l2": 2.5091508929606365, | |
| "delta_retention_norm_ratio": 0.874340750844961, | |
| "delta_cosine": 0.6862452411628439, | |
| "relative_weight_rmse": 0.0120082794787516 | |
| }, | |
| { | |
| "index": 52, | |
| "total": 252, | |
| "projection": "layers.15.mlp.up_proj", | |
| "shape": [ | |
| 12288, | |
| 4096 | |
| ], | |
| "seconds": 0.3788, | |
| "parameters": 50331648, | |
| "reference_delta_l2": 2.7874161707431244, | |
| "retained_delta_l2": 2.438701622275863, | |
| "delta_retention_norm_ratio": 0.874896848153717, | |
| "delta_cosine": 0.689275565248844, | |
| "relative_weight_rmse": 0.011420313770719347 | |
| }, | |
| { | |
| "index": 53, | |
| "total": 252, | |
| "projection": "layers.15.self_attn.k_proj", | |
| "shape": [ | |
| 1024, | |
| 4096 | |
| ], | |
| "seconds": 0.0304, | |
| "parameters": 4194304, | |
| "reference_delta_l2": 0.8128830019881568, | |
| "retained_delta_l2": 0.7186374562350791, | |
| "delta_retention_norm_ratio": 0.8840601347025697, | |
| "delta_cosine": 0.6988000834000759, | |
| "relative_weight_rmse": 0.011576737232218115 | |
| }, | |
| { | |
| "index": 54, | |
| "total": 252, | |
| "projection": "layers.15.self_attn.o_proj", | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "seconds": 0.1271, | |
| "parameters": 16777216, | |
| "reference_delta_l2": 2.150099775903978, | |
| "retained_delta_l2": 2.0663030243918636, | |
| "delta_retention_norm_ratio": 0.9610265753937475, | |
| "delta_cosine": 0.7885854934462535, | |
| "relative_weight_rmse": 0.013839239409670728 | |
| }, | |
| { | |
| "index": 55, | |
| "total": 252, | |
| "projection": "layers.15.self_attn.q_proj", | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "seconds": 0.1272, | |
| "parameters": 16777216, | |
| "reference_delta_l2": 2.2362415876696247, | |
| "retained_delta_l2": 2.1611246850921555, | |
| "delta_retention_norm_ratio": 0.9664093079246648, | |
| "delta_cosine": 0.801026727693609, | |
| "relative_weight_rmse": 0.013447972564126687 | |
| }, | |
| { | |
| "index": 56, | |
| "total": 252, | |
| "projection": "layers.15.self_attn.v_proj", | |
| "shape": [ | |
| 1024, | |
| 4096 | |
| ], | |
| "seconds": 0.029, | |
| "parameters": 4194304, | |
| "reference_delta_l2": 0.855355924267693, | |
| "retained_delta_l2": 0.7597572597045174, | |
| "delta_retention_norm_ratio": 0.8882352224952182, | |
| "delta_cosine": 0.704667737659905, | |
| "relative_weight_rmse": 0.011198650304465709 | |
| }, | |
| { | |
| "index": 57, | |
| "total": 252, | |
| "projection": "layers.16.mlp.down_proj", | |
| "shape": [ | |
| 4096, | |
| 12288 | |
| ], | |
| "seconds": 0.4184, | |
| "parameters": 50331648, | |
| "reference_delta_l2": 2.490608050723674, | |
| "retained_delta_l2": 2.1276880125739366, | |
| "delta_retention_norm_ratio": 0.8542845639464277, | |
| "delta_cosine": 0.6690464044195373, | |
| "relative_weight_rmse": 0.010663801946869307 | |
| }, | |
| { | |
| "index": 58, | |
| "total": 252, | |
| "projection": "layers.16.mlp.gate_proj", | |
| "shape": [ | |
| 12288, | |
| 4096 | |
| ], | |
| "seconds": 0.3881, | |
| "parameters": 50331648, | |
| "reference_delta_l2": 2.6372762406162753, | |
| "retained_delta_l2": 2.253994849358787, | |
| "delta_retention_norm_ratio": 0.854667711575059, | |
| "delta_cosine": 0.6681974156909103, | |
| "relative_weight_rmse": 0.011063871539140003 | |
| }, | |
| { | |
| "index": 59, | |
| "total": 252, | |
| "projection": "layers.16.mlp.up_proj", | |
| "shape": [ | |
| 12288, | |
| 4096 | |
| ], | |
| "seconds": 0.38, | |
| "parameters": 50331648, | |
| "reference_delta_l2": 2.5562915910917727, | |
| "retained_delta_l2": 2.154870300649257, | |
| "delta_retention_norm_ratio": 0.8429673313320756, | |
| "delta_cosine": 0.6570978221314555, | |
| "relative_weight_rmse": 0.01088591802268865 | |
| }, | |
| { | |
| "index": 60, | |
| "total": 252, | |
| "projection": "layers.16.self_attn.k_proj", | |
| "shape": [ | |
| 1024, | |
| 4096 | |
| ], | |
| "seconds": 0.0307, | |
| "parameters": 4194304, | |
| "reference_delta_l2": 0.7402672696675293, | |
| "retained_delta_l2": 0.6376569454010689, | |
| "delta_retention_norm_ratio": 0.8613874630543302, | |
| "delta_cosine": 0.675448610838514, | |
| "relative_weight_rmse": 0.011215528271285295 | |
| }, | |
| { | |
| "index": 61, | |
| "total": 252, | |
| "projection": "layers.16.self_attn.o_proj", | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "seconds": 0.1304, | |
| "parameters": 16777216, | |
| "reference_delta_l2": 1.6665277447725417, | |
| "retained_delta_l2": 1.4904262957381975, | |
| "delta_retention_norm_ratio": 0.8943303226803586, | |
| "delta_cosine": 0.7094085759432628, | |
| "relative_weight_rmse": 0.01193810158565392 | |
| }, | |
| { | |
| "index": 62, | |
| "total": 252, | |
| "projection": "layers.16.self_attn.q_proj", | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "seconds": 0.1275, | |
| "parameters": 16777216, | |
| "reference_delta_l2": 1.8094508759468877, | |
| "retained_delta_l2": 1.677429719498606, | |
| "delta_retention_norm_ratio": 0.9270379990950598, | |
| "delta_cosine": 0.7479558975300578, | |
| "relative_weight_rmse": 0.012459277798183784 | |
| }, | |
| { | |
| "index": 63, | |
| "total": 252, | |
| "projection": "layers.16.self_attn.v_proj", | |
| "shape": [ | |
| 1024, | |
| 4096 | |
| ], | |
| "seconds": 0.0295, | |
| "parameters": 4194304, | |
| "reference_delta_l2": 0.9588999684784266, | |
| "retained_delta_l2": 0.8758522584149606, | |
| "delta_retention_norm_ratio": 0.9133927283413668, | |
| "delta_cosine": 0.7302107892117504, | |
| "relative_weight_rmse": 0.012150515736783439 | |
| }, | |
| { | |
| "index": 64, | |
| "total": 252, | |
| "projection": "layers.17.mlp.down_proj", | |
| "shape": [ | |
| 4096, | |
| 12288 | |
| ], | |
| "seconds": 0.4182, | |
| "parameters": 50331648, | |
| "reference_delta_l2": 2.57867394869486, | |
| "retained_delta_l2": 2.217903125991572, | |
| "delta_retention_norm_ratio": 0.8600944400567259, | |
| "delta_cosine": 0.6745827781840225, | |
| "relative_weight_rmse": 0.010922727095506776 | |
| }, | |
| { | |
| "index": 65, | |
| "total": 252, | |
| "projection": "layers.17.mlp.gate_proj", | |
| "shape": [ | |
| 12288, | |
| 4096 | |
| ], | |
| "seconds": 0.3773, | |
| "parameters": 50331648, | |
| "reference_delta_l2": 3.530974860487244, | |
| "retained_delta_l2": 3.309138766374794, | |
| "delta_retention_norm_ratio": 0.9371742640835912, | |
| "delta_cosine": 0.7559309585248899, | |
| "relative_weight_rmse": 0.013279111898597819 | |
| }, | |
| { | |
| "index": 66, | |
| "total": 252, | |
| "projection": "layers.17.mlp.up_proj", | |
| "shape": [ | |
| 12288, | |
| 4096 | |
| ], | |
| "seconds": 0.3772, | |
| "parameters": 50331648, | |
| "reference_delta_l2": 3.182246137690255, | |
| "retained_delta_l2": 2.8980044926585204, | |
| "delta_retention_norm_ratio": 0.9106789252832457, | |
| "delta_cosine": 0.725261046959691, | |
| "relative_weight_rmse": 0.012372349212028403 | |
| }, | |
| { | |
| "index": 67, | |
| "total": 252, | |
| "projection": "layers.17.self_attn.k_proj", | |
| "shape": [ | |
| 1024, | |
| 4096 | |
| ], | |
| "seconds": 0.0306, | |
| "parameters": 4194304, | |
| "reference_delta_l2": 0.7856562521842315, | |
| "retained_delta_l2": 0.7028263759527129, | |
| "delta_retention_norm_ratio": 0.8945723705485189, | |
| "delta_cosine": 0.7142405090358148, | |
| "relative_weight_rmse": 0.011541983384061129 | |
| }, | |
| { | |
| "index": 68, | |
| "total": 252, | |
| "projection": "layers.17.self_attn.o_proj", | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "seconds": 0.1281, | |
| "parameters": 16777216, | |
| "reference_delta_l2": 1.7288860895233422, | |
| "retained_delta_l2": 1.5687744025872319, | |
| "delta_retention_norm_ratio": 0.9073902624896164, | |
| "delta_cosine": 0.7245259556841699, | |
| "relative_weight_rmse": 0.012197539719324683 | |
| }, | |
| { | |
| "index": 69, | |
| "total": 252, | |
| "projection": "layers.17.self_attn.q_proj", | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "seconds": 0.1276, | |
| "parameters": 16777216, | |
| "reference_delta_l2": 1.659428444075962, | |
| "retained_delta_l2": 1.4844032143221468, | |
| "delta_retention_norm_ratio": 0.8945267990441875, | |
| "delta_cosine": 0.7074773036620385, | |
| "relative_weight_rmse": 0.011963215771788813 | |
| }, | |
| { | |
| "index": 70, | |
| "total": 252, | |
| "projection": "layers.17.self_attn.v_proj", | |
| "shape": [ | |
| 1024, | |
| 4096 | |
| ], | |
| "seconds": 0.0294, | |
| "parameters": 4194304, | |
| "reference_delta_l2": 1.033762917861183, | |
| "retained_delta_l2": 0.9695695450654802, | |
| "delta_retention_norm_ratio": 0.9379031964809529, | |
| "delta_cosine": 0.7590274849022801, | |
| "relative_weight_rmse": 0.012764403414386711 | |
| }, | |
| { | |
| "index": 71, | |
| "total": 252, | |
| "projection": "layers.18.mlp.down_proj", | |
| "shape": [ | |
| 4096, | |
| 12288 | |
| ], | |
| "seconds": 0.4196, | |
| "parameters": 50331648, | |
| "reference_delta_l2": 2.910789276888511, | |
| "retained_delta_l2": 2.6222802085342862, | |
| "delta_retention_norm_ratio": 0.9008828737123057, | |
| "delta_cosine": 0.7154197570893368, | |
| "relative_weight_rmse": 0.011715650325460434 | |
| }, | |
| { | |
| "index": 72, | |
| "total": 252, | |
| "projection": "layers.18.mlp.gate_proj", | |
| "shape": [ | |
| 12288, | |
| 4096 | |
| ], | |
| "seconds": 0.3795, | |
| "parameters": 50331648, | |
| "reference_delta_l2": 4.722324488785237, | |
| "retained_delta_l2": 4.68356445391355, | |
| "delta_retention_norm_ratio": 0.9917921703678483, | |
| "delta_cosine": 0.8270502288163069, | |
| "relative_weight_rmse": 0.015419870967505798 | |
| }, | |
| { | |
| "index": 73, | |
| "total": 252, | |
| "projection": "layers.18.mlp.up_proj", | |
| "shape": [ | |
| 12288, | |
| 4096 | |
| ], | |
| "seconds": 0.3803, | |
| "parameters": 50331648, | |
| "reference_delta_l2": 3.267632846087272, | |
| "retained_delta_l2": 3.0065344734979704, | |
| "delta_retention_norm_ratio": 0.9200955600314932, | |
| "delta_cosine": 0.7366907817166277, | |
| "relative_weight_rmse": 0.012487845832375657 | |
| }, | |
| { | |
| "index": 74, | |
| "total": 252, | |
| "projection": "layers.18.self_attn.k_proj", | |
| "shape": [ | |
| 1024, | |
| 4096 | |
| ], | |
| "seconds": 0.0304, | |
| "parameters": 4194304, | |
| "reference_delta_l2": 1.1178209584979375, | |
| "retained_delta_l2": 1.0843822621846049, | |
| "delta_retention_norm_ratio": 0.9700858209365965, | |
| "delta_cosine": 0.7990006485228375, | |
| "relative_weight_rmse": 0.014000429949694236 | |
| }, | |
| { | |
| "index": 75, | |
| "total": 252, | |
| "projection": "layers.18.self_attn.o_proj", | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "seconds": 0.1282, | |
| "parameters": 16777216, | |
| "reference_delta_l2": 1.6294500157772767, | |
| "retained_delta_l2": 1.4565664018421354, | |
| "delta_retention_norm_ratio": 0.8939006337959542, | |
| "delta_cosine": 0.7108714273975982, | |
| "relative_weight_rmse": 0.011701311661202553 | |
| }, | |
| { | |
| "index": 76, | |
| "total": 252, | |
| "projection": "layers.18.self_attn.q_proj", | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "seconds": 0.1303, | |
| "parameters": 16777216, | |
| "reference_delta_l2": 2.099352970985407, | |
| "retained_delta_l2": 2.009225849712284, | |
| "delta_retention_norm_ratio": 0.957069095802971, | |
| "delta_cosine": 0.7836883856589295, | |
| "relative_weight_rmse": 0.013542395647038019 | |
| }, | |
| { | |
| "index": 77, | |
| "total": 252, | |
| "projection": "layers.18.self_attn.v_proj", | |
| "shape": [ | |
| 1024, | |
| 4096 | |
| ], | |
| "seconds": 0.0291, | |
| "parameters": 4194304, | |
| "reference_delta_l2": 0.9281397272835645, | |
| "retained_delta_l2": 0.8453631539465025, | |
| "delta_retention_norm_ratio": 0.9108145348122006, | |
| "delta_cosine": 0.7279251236491474, | |
| "relative_weight_rmse": 0.011786639770044518 | |
| }, | |
| { | |
| "index": 78, | |
| "total": 252, | |
| "projection": "layers.19.mlp.down_proj", | |
| "shape": [ | |
| 4096, | |
| 12288 | |
| ], | |
| "seconds": 0.4218, | |
| "parameters": 50331648, | |
| "reference_delta_l2": 2.6906724529290753, | |
| "retained_delta_l2": 2.353773769212275, | |
| "delta_retention_norm_ratio": 0.8747901539074179, | |
| "delta_cosine": 0.6890261111788295, | |
| "relative_weight_rmse": 0.011201229115173 | |
| }, | |
| { | |
| "index": 79, | |
| "total": 252, | |
| "projection": "layers.19.mlp.gate_proj", | |
| "shape": [ | |
| 12288, | |
| 4096 | |
| ], | |
| "seconds": 0.379, | |
| "parameters": 50331648, | |
| "reference_delta_l2": 4.6011590568420395, | |
| "retained_delta_l2": 4.5324392054321665, | |
| "delta_retention_norm_ratio": 0.9850646651070049, | |
| "delta_cosine": 0.8179573786954442, | |
| "relative_weight_rmse": 0.015420800468068915 | |
| }, | |
| { | |
| "index": 80, | |
| "total": 252, | |
| "projection": "layers.19.mlp.up_proj", | |
| "shape": [ | |
| 12288, | |
| 4096 | |
| ], | |
| "seconds": 0.3787, | |
| "parameters": 50331648, | |
| "reference_delta_l2": 3.088979201522636, | |
| "retained_delta_l2": 2.7691085642458395, | |
| "delta_retention_norm_ratio": 0.8964477853657532, | |
| "delta_cosine": 0.7106646889590305, | |
| "relative_weight_rmse": 0.012187361703435885 | |
| }, | |
| { | |
| "index": 81, | |
| "total": 252, | |
| "projection": "layers.19.self_attn.k_proj", | |
| "shape": [ | |
| 1024, | |
| 4096 | |
| ], | |
| "seconds": 0.0305, | |
| "parameters": 4194304, | |
| "reference_delta_l2": 1.0424679931914167, | |
| "retained_delta_l2": 1.0031846977199295, | |
| "delta_retention_norm_ratio": 0.9623170248601829, | |
| "delta_cosine": 0.7948003347645736, | |
| "relative_weight_rmse": 0.01357524445006692 | |
| }, | |
| { | |
| "index": 82, | |
| "total": 252, | |
| "projection": "layers.19.self_attn.o_proj", | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "seconds": 0.1285, | |
| "parameters": 16777216, | |
| "reference_delta_l2": 1.6393473573479356, | |
| "retained_delta_l2": 1.4597608928780879, | |
| "delta_retention_norm_ratio": 0.8904524634971964, | |
| "delta_cosine": 0.7051363882562122, | |
| "relative_weight_rmse": 0.011792228026413542 | |
| }, | |
| { | |
| "index": 83, | |
| "total": 252, | |
| "projection": "layers.19.self_attn.q_proj", | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "seconds": 0.128, | |
| "parameters": 16777216, | |
| "reference_delta_l2": 1.9324947526748775, | |
| "retained_delta_l2": 1.8224312290030995, | |
| "delta_retention_norm_ratio": 0.9430458874367277, | |
| "delta_cosine": 0.765239868235387, | |
| "relative_weight_rmse": 0.012807740543361366 | |
| }, | |
| { | |
| "index": 84, | |
| "total": 252, | |
| "projection": "layers.19.self_attn.v_proj", | |
| "shape": [ | |
| 1024, | |
| 4096 | |
| ], | |
| "seconds": 0.0294, | |
| "parameters": 4194304, | |
| "reference_delta_l2": 1.2393259702908999, | |
| "retained_delta_l2": 1.2060395661274452, | |
| "delta_retention_norm_ratio": 0.973141526150992, | |
| "delta_cosine": 0.806109674730439, | |
| "relative_weight_rmse": 0.013900123324460853 | |
| }, | |
| { | |
| "index": 85, | |
| "total": 252, | |
| "projection": "layers.2.mlp.down_proj", | |
| "shape": [ | |
| 4096, | |
| 12288 | |
| ], | |
| "seconds": 0.4212, | |
| "parameters": 50331648, | |
| "reference_delta_l2": 1.550514040883994, | |
| "retained_delta_l2": 1.3354703881576557, | |
| "delta_retention_norm_ratio": 0.8613081551949471, | |
| "delta_cosine": 0.6980816143317669, | |
| "relative_weight_rmse": 0.010102768254166766 | |
| }, | |
| { | |
| "index": 86, | |
| "total": 252, | |
| "projection": "layers.2.mlp.gate_proj", | |
| "shape": [ | |
| 12288, | |
| 4096 | |
| ], | |
| "seconds": 0.3786, | |
| "parameters": 50331648, | |
| "reference_delta_l2": 2.5024354983712715, | |
| "retained_delta_l2": 2.3585423713079154, | |
| "delta_retention_norm_ratio": 0.942498766838542, | |
| "delta_cosine": 0.7967290998301102, | |
| "relative_weight_rmse": 0.012645736077248214 | |
| }, | |
| { | |
| "index": 87, | |
| "total": 252, | |
| "projection": "layers.2.mlp.up_proj", | |
| "shape": [ | |
| 12288, | |
| 4096 | |
| ], | |
| "seconds": 0.3774, | |
| "parameters": 50331648, | |
| "reference_delta_l2": 3.369649175371511, | |
| "retained_delta_l2": 3.3497295641104907, | |
| "delta_retention_norm_ratio": 0.9940885207259524, | |
| "delta_cosine": 0.8793330738810422, | |
| "relative_weight_rmse": 0.01495990024596177 | |
| }, | |
| { | |
| "index": 88, | |
| "total": 252, | |
| "projection": "layers.2.self_attn.k_proj", | |
| "shape": [ | |
| 1024, | |
| 4096 | |
| ], | |
| "seconds": 0.0305, | |
| "parameters": 4194304, | |
| "reference_delta_l2": 0.5772932706105626, | |
| "retained_delta_l2": 0.4344806465765585, | |
| "delta_retention_norm_ratio": 0.7526168564498228, | |
| "delta_cosine": 0.5763063069305538, | |
| "relative_weight_rmse": 0.009291145467617177 | |
| }, | |
| { | |
| "index": 89, | |
| "total": 252, | |
| "projection": "layers.2.self_attn.o_proj", | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "seconds": 0.1283, | |
| "parameters": 16777216, | |
| "reference_delta_l2": 2.0861834258535827, | |
| "retained_delta_l2": 1.9678414833336078, | |
| "delta_retention_norm_ratio": 0.9432734719999254, | |
| "delta_cosine": 0.7605381226137655, | |
| "relative_weight_rmse": 0.013949476644897577 | |
| }, | |
| { | |
| "index": 90, | |
| "total": 252, | |
| "projection": "layers.2.self_attn.q_proj", | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "seconds": 0.1273, | |
| "parameters": 16777216, | |
| "reference_delta_l2": 1.2282960127488745, | |
| "retained_delta_l2": 0.9700631471429637, | |
| "delta_retention_norm_ratio": 0.7897633282811066, | |
| "delta_cosine": 0.6079123164782136, | |
| "relative_weight_rmse": 0.010177397449548601 | |
| }, | |
| { | |
| "index": 91, | |
| "total": 252, | |
| "projection": "layers.2.self_attn.v_proj", | |
| "shape": [ | |
| 1024, | |
| 4096 | |
| ], | |
| "seconds": 0.03, | |
| "parameters": 4194304, | |
| "reference_delta_l2": 1.1607808521417493, | |
| "retained_delta_l2": 1.1195868281071086, | |
| "delta_retention_norm_ratio": 0.9645117991405234, | |
| "delta_cosine": 0.793185053055849, | |
| "relative_weight_rmse": 0.01351105097316285 | |
| }, | |
| { | |
| "index": 92, | |
| "total": 252, | |
| "projection": "layers.20.mlp.down_proj", | |
| "shape": [ | |
| 4096, | |
| 12288 | |
| ], | |
| "seconds": 0.4246, | |
| "parameters": 50331648, | |
| "reference_delta_l2": 2.64584748451423, | |
| "retained_delta_l2": 2.258222891237714, | |
| "delta_retention_norm_ratio": 0.8534970002824321, | |
| "delta_cosine": 0.6658774979676224, | |
| "relative_weight_rmse": 0.011247267660759373 | |
| }, | |
| { | |
| "index": 93, | |
| "total": 252, | |
| "projection": "layers.20.mlp.gate_proj", | |
| "shape": [ | |
| 12288, | |
| 4096 | |
| ], | |
| "seconds": 0.3789, | |
| "parameters": 50331648, | |
| "reference_delta_l2": 4.526615232031967, | |
| "retained_delta_l2": 4.455530846754728, | |
| "delta_retention_norm_ratio": 0.984296349118825, | |
| "delta_cosine": 0.8160620069203229, | |
| "relative_weight_rmse": 0.01538652110350443 | |
| }, | |
| { | |
| "index": 94, | |
| "total": 252, | |
| "projection": "layers.20.mlp.up_proj", | |
| "shape": [ | |
| 12288, | |
| 4096 | |
| ], | |
| "seconds": 0.3776, | |
| "parameters": 50331648, | |
| "reference_delta_l2": 3.1237101903647013, | |
| "retained_delta_l2": 2.806044559157967, | |
| "delta_retention_norm_ratio": 0.8983050245228908, | |
| "delta_cosine": 0.7120996555848715, | |
| "relative_weight_rmse": 0.012270615442135124 | |
| }, | |
| { | |
| "index": 95, | |
| "total": 252, | |
| "projection": "layers.20.self_attn.k_proj", | |
| "shape": [ | |
| 1024, | |
| 4096 | |
| ], | |
| "seconds": 0.0305, | |
| "parameters": 4194304, | |
| "reference_delta_l2": 0.8431576555668275, | |
| "retained_delta_l2": 0.7751642730249042, | |
| "delta_retention_norm_ratio": 0.9193586370318685, | |
| "delta_cosine": 0.739759292876534, | |
| "relative_weight_rmse": 0.012192669685611647 | |
| }, | |
| { | |
| "index": 96, | |
| "total": 252, | |
| "projection": "layers.20.self_attn.o_proj", | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "seconds": 0.1283, | |
| "parameters": 16777216, | |
| "reference_delta_l2": 1.6098180721363975, | |
| "retained_delta_l2": 1.428732374756382, | |
| "delta_retention_norm_ratio": 0.8875117005366353, | |
| "delta_cosine": 0.7034375530109004, | |
| "relative_weight_rmse": 0.011826447278611532 | |
| }, | |
| { | |
| "index": 97, | |
| "total": 252, | |
| "projection": "layers.20.self_attn.q_proj", | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "seconds": 0.128, | |
| "parameters": 16777216, | |
| "reference_delta_l2": 1.8915653169770341, | |
| "retained_delta_l2": 1.7676069670292385, | |
| "delta_retention_norm_ratio": 0.9344678458442571, | |
| "delta_cosine": 0.7568589852551876, | |
| "relative_weight_rmse": 0.012334276756101584 | |
| }, | |
| { | |
| "index": 98, | |
| "total": 252, | |
| "projection": "layers.20.self_attn.v_proj", | |
| "shape": [ | |
| 1024, | |
| 4096 | |
| ], | |
| "seconds": 0.0294, | |
| "parameters": 4194304, | |
| "reference_delta_l2": 0.9440685914528588, | |
| "retained_delta_l2": 0.8662821368972096, | |
| "delta_retention_norm_ratio": 0.9176050816011778, | |
| "delta_cosine": 0.7369546341937735, | |
| "relative_weight_rmse": 0.012020241368954576 | |
| }, | |
| { | |
| "index": 99, | |
| "total": 252, | |
| "projection": "layers.21.mlp.down_proj", | |
| "shape": [ | |
| 4096, | |
| 12288 | |
| ], | |
| "seconds": 0.4276, | |
| "parameters": 50331648, | |
| "reference_delta_l2": 3.105774206302042, | |
| "retained_delta_l2": 2.8037671599278715, | |
| "delta_retention_norm_ratio": 0.9027594968876499, | |
| "delta_cosine": 0.7157717939258454, | |
| "relative_weight_rmse": 0.012307599987023648 | |
| }, | |
| { | |
| "index": 100, | |
| "total": 252, | |
| "projection": "layers.21.mlp.gate_proj", | |
| "shape": [ | |
| 12288, | |
| 4096 | |
| ], | |
| "seconds": 0.3775, | |
| "parameters": 50331648, | |
| "reference_delta_l2": 4.603324023865813, | |
| "retained_delta_l2": 4.538266389340112, | |
| "delta_retention_norm_ratio": 0.9858672484951284, | |
| "delta_cosine": 0.8160356108217762, | |
| "relative_weight_rmse": 0.015878057128781025 | |
| }, | |
| { | |
| "index": 101, | |
| "total": 252, | |
| "projection": "layers.21.mlp.up_proj", | |
| "shape": [ | |
| 12288, | |
| 4096 | |
| ], | |
| "seconds": 0.3776, | |
| "parameters": 50331648, | |
| "reference_delta_l2": 3.339274323389295, | |
| "retained_delta_l2": 3.0655218952248084, | |
| "delta_retention_norm_ratio": 0.9180203835764432, | |
| "delta_cosine": 0.7329672214327478, | |
| "relative_weight_rmse": 0.012754384253425642 | |
| }, | |
| { | |
| "index": 102, | |
| "total": 252, | |
| "projection": "layers.21.self_attn.k_proj", | |
| "shape": [ | |
| 1024, | |
| 4096 | |
| ], | |
| "seconds": 0.0305, | |
| "parameters": 4194304, | |
| "reference_delta_l2": 0.8880355259681696, | |
| "retained_delta_l2": 0.8209643352818649, | |
| "delta_retention_norm_ratio": 0.9244724014693204, | |
| "delta_cosine": 0.7418171289261559, | |
| "relative_weight_rmse": 0.01281140542251685 | |
| }, | |
| { | |
| "index": 103, | |
| "total": 252, | |
| "projection": "layers.21.self_attn.o_proj", | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "seconds": 0.1279, | |
| "parameters": 16777216, | |
| "reference_delta_l2": 1.7371948241009854, | |
| "retained_delta_l2": 1.58872774854642, | |
| "delta_retention_norm_ratio": 0.9145363125109479, | |
| "delta_cosine": 0.7322945496835428, | |
| "relative_weight_rmse": 0.012386180007419076 | |
| }, | |
| { | |
| "index": 104, | |
| "total": 252, | |
| "projection": "layers.21.self_attn.q_proj", | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "seconds": 0.1263, | |
| "parameters": 16777216, | |
| "reference_delta_l2": 1.345556330912787, | |
| "retained_delta_l2": 1.1124933541307664, | |
| "delta_retention_norm_ratio": 0.8267906207806868, | |
| "delta_cosine": 0.6435156791814104, | |
| "relative_weight_rmse": 0.010325715461482674 | |
| }, | |
| { | |
| "index": 105, | |
| "total": 252, | |
| "projection": "layers.21.self_attn.v_proj", | |
| "shape": [ | |
| 1024, | |
| 4096 | |
| ], | |
| "seconds": 0.0295, | |
| "parameters": 4194304, | |
| "reference_delta_l2": 1.3430392811254306, | |
| "retained_delta_l2": 1.3264946568200315, | |
| "delta_retention_norm_ratio": 0.9876812059498847, | |
| "delta_cosine": 0.8236209766657606, | |
| "relative_weight_rmse": 0.014574284287276256 | |
| }, | |
| { | |
| "index": 106, | |
| "total": 252, | |
| "projection": "layers.22.mlp.down_proj", | |
| "shape": [ | |
| 4096, | |
| 12288 | |
| ], | |
| "seconds": 0.4185, | |
| "parameters": 50331648, | |
| "reference_delta_l2": 3.3563104342236816, | |
| "retained_delta_l2": 3.1200426688610725, | |
| "delta_retention_norm_ratio": 0.9296049128967839, | |
| "delta_cosine": 0.7472098403303402, | |
| "relative_weight_rmse": 0.012595485823080576 | |
| }, | |
| { | |
| "index": 107, | |
| "total": 252, | |
| "projection": "layers.22.mlp.gate_proj", | |
| "shape": [ | |
| 12288, | |
| 4096 | |
| ], | |
| "seconds": 0.377, | |
| "parameters": 50331648, | |
| "reference_delta_l2": 4.207905625812568, | |
| "retained_delta_l2": 4.092846670489135, | |
| "delta_retention_norm_ratio": 0.9726564791240501, | |
| "delta_cosine": 0.8018039373008636, | |
| "relative_weight_rmse": 0.01464523430893868 | |
| }, | |
| { | |
| "index": 108, | |
| "total": 252, | |
| "projection": "layers.22.mlp.up_proj", | |
| "shape": [ | |
| 12288, | |
| 4096 | |
| ], | |
| "seconds": 0.3767, | |
| "parameters": 50331648, | |
| "reference_delta_l2": 3.4246106949522344, | |
| "retained_delta_l2": 3.173732205832326, | |
| "delta_retention_norm_ratio": 0.926742479228458, | |
| "delta_cosine": 0.7448358030341105, | |
| "relative_weight_rmse": 0.01261764380333146 | |
| }, | |
| { | |
| "index": 109, | |
| "total": 252, | |
| "projection": "layers.22.self_attn.k_proj", | |
| "shape": [ | |
| 1024, | |
| 4096 | |
| ], | |
| "seconds": 0.0303, | |
| "parameters": 4194304, | |
| "reference_delta_l2": 1.1250380548192516, | |
| "retained_delta_l2": 1.099810524114935, | |
| "delta_retention_norm_ratio": 0.9775762867786996, | |
| "delta_cosine": 0.8176936058321127, | |
| "relative_weight_rmse": 0.01441093779238604 | |
| }, | |
| { | |
| "index": 110, | |
| "total": 252, | |
| "projection": "layers.22.self_attn.o_proj", | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "seconds": 0.1272, | |
| "parameters": 16777216, | |
| "reference_delta_l2": 2.0979665245617816, | |
| "retained_delta_l2": 1.9974504967278541, | |
| "delta_retention_norm_ratio": 0.9520888314197848, | |
| "delta_cosine": 0.7727993917982432, | |
| "relative_weight_rmse": 0.013730056442512895 | |
| }, | |
| { | |
| "index": 111, | |
| "total": 252, | |
| "projection": "layers.22.self_attn.q_proj", | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "seconds": 0.1284, | |
| "parameters": 16777216, | |
| "reference_delta_l2": 1.869767544241714, | |
| "retained_delta_l2": 1.7405896445630222, | |
| "delta_retention_norm_ratio": 0.9309123211189978, | |
| "delta_cosine": 0.7546608977847129, | |
| "relative_weight_rmse": 0.01259283284123554 | |
| }, | |
| { | |
| "index": 112, | |
| "total": 252, | |
| "projection": "layers.22.self_attn.v_proj", | |
| "shape": [ | |
| 1024, | |
| 4096 | |
| ], | |
| "seconds": 0.0298, | |
| "parameters": 4194304, | |
| "reference_delta_l2": 1.1037279157423996, | |
| "retained_delta_l2": 1.0516996683747628, | |
| "delta_retention_norm_ratio": 0.9528613468722127, | |
| "delta_cosine": 0.77769661979401, | |
| "relative_weight_rmse": 0.013220168147444164 | |
| }, | |
| { | |
| "index": 113, | |
| "total": 252, | |
| "projection": "layers.23.mlp.down_proj", | |
| "shape": [ | |
| 4096, | |
| 12288 | |
| ], | |
| "seconds": 0.4216, | |
| "parameters": 50331648, | |
| "reference_delta_l2": 3.160235840871493, | |
| "retained_delta_l2": 2.9006612353003223, | |
| "delta_retention_norm_ratio": 0.9178622676782286, | |
| "delta_cosine": 0.7351433392668709, | |
| "relative_weight_rmse": 0.011972793198786804 | |
| }, | |
| { | |
| "index": 114, | |
| "total": 252, | |
| "projection": "layers.23.mlp.gate_proj", | |
| "shape": [ | |
| 12288, | |
| 4096 | |
| ], | |
| "seconds": 0.3778, | |
| "parameters": 50331648, | |
| "reference_delta_l2": 5.626287078192542, | |
| "retained_delta_l2": 5.669300174595202, | |
| "delta_retention_norm_ratio": 1.007645023406178, | |
| "delta_cosine": 0.8530969607907128, | |
| "relative_weight_rmse": 0.01679839814486996 | |
| }, | |
| { | |
| "index": 115, | |
| "total": 252, | |
| "projection": "layers.23.mlp.up_proj", | |
| "shape": [ | |
| 12288, | |
| 4096 | |
| ], | |
| "seconds": 0.3775, | |
| "parameters": 50331648, | |
| "reference_delta_l2": 4.980546573065696, | |
| "retained_delta_l2": 4.943087972922342, | |
| "delta_retention_norm_ratio": 0.9924790181973349, | |
| "delta_cosine": 0.8288916075112348, | |
| "relative_weight_rmse": 0.015410818804014629 | |
| }, | |
| { | |
| "index": 116, | |
| "total": 252, | |
| "projection": "layers.23.self_attn.k_proj", | |
| "shape": [ | |
| 1024, | |
| 4096 | |
| ], | |
| "seconds": 0.0301, | |
| "parameters": 4194304, | |
| "reference_delta_l2": 0.7941298246634246, | |
| "retained_delta_l2": 0.7066560415227385, | |
| "delta_retention_norm_ratio": 0.8898495177690121, | |
| "delta_cosine": 0.7054838997759174, | |
| "relative_weight_rmse": 0.011970656431982295 | |
| }, | |
| { | |
| "index": 117, | |
| "total": 252, | |
| "projection": "layers.23.self_attn.o_proj", | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "seconds": 0.1327, | |
| "parameters": 16777216, | |
| "reference_delta_l2": 1.9616796991461982, | |
| "retained_delta_l2": 1.834908647487434, | |
| "delta_retention_norm_ratio": 0.9353762738565629, | |
| "delta_cosine": 0.7548535319766523, | |
| "relative_weight_rmse": 0.012811803614575604 | |
| }, | |
| { | |
| "index": 118, | |
| "total": 252, | |
| "projection": "layers.23.self_attn.q_proj", | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "seconds": 0.1304, | |
| "parameters": 16777216, | |
| "reference_delta_l2": 1.7868156453388941, | |
| "retained_delta_l2": 1.6194093744676163, | |
| "delta_retention_norm_ratio": 0.9063102725186141, | |
| "delta_cosine": 0.7240533048699785, | |
| "relative_weight_rmse": 0.01202873511087103 | |
| }, | |
| { | |
| "index": 119, | |
| "total": 252, | |
| "projection": "layers.23.self_attn.v_proj", | |
| "shape": [ | |
| 1024, | |
| 4096 | |
| ], | |
| "seconds": 0.0293, | |
| "parameters": 4194304, | |
| "reference_delta_l2": 0.9269632845956063, | |
| "retained_delta_l2": 0.8512550950173756, | |
| "delta_retention_norm_ratio": 0.9183266577691274, | |
| "delta_cosine": 0.7375855337175854, | |
| "relative_weight_rmse": 0.011726307762073923 | |
| }, | |
| { | |
| "index": 120, | |
| "total": 252, | |
| "projection": "layers.24.mlp.down_proj", | |
| "shape": [ | |
| 4096, | |
| 12288 | |
| ], | |
| "seconds": 0.4183, | |
| "parameters": 50331648, | |
| "reference_delta_l2": 3.2514706960498443, | |
| "retained_delta_l2": 2.9793935187526754, | |
| "delta_retention_norm_ratio": 0.9163218116565803, | |
| "delta_cosine": 0.7316018859300564, | |
| "relative_weight_rmse": 0.01227576206795573 | |
| }, | |
| { | |
| "index": 121, | |
| "total": 252, | |
| "projection": "layers.24.mlp.gate_proj", | |
| "shape": [ | |
| 12288, | |
| 4096 | |
| ], | |
| "seconds": 0.3792, | |
| "parameters": 50331648, | |
| "reference_delta_l2": 5.11220125208371, | |
| "retained_delta_l2": 5.0793680797612115, | |
| "delta_retention_norm_ratio": 0.9935774883062133, | |
| "delta_cosine": 0.8268587806603359, | |
| "relative_weight_rmse": 0.01622681210023664 | |
| }, | |
| { | |
| "index": 122, | |
| "total": 252, | |
| "projection": "layers.24.mlp.up_proj", | |
| "shape": [ | |
| 12288, | |
| 4096 | |
| ], | |
| "seconds": 0.3823, | |
| "parameters": 50331648, | |
| "reference_delta_l2": 4.008939223739621, | |
| "retained_delta_l2": 3.8306350545945733, | |
| "delta_retention_norm_ratio": 0.9555233543853223, | |
| "delta_cosine": 0.7755895473310437, | |
| "relative_weight_rmse": 0.013947869007459165 | |
| }, | |
| { | |
| "index": 123, | |
| "total": 252, | |
| "projection": "layers.24.self_attn.k_proj", | |
| "shape": [ | |
| 1024, | |
| 4096 | |
| ], | |
| "seconds": 0.0306, | |
| "parameters": 4194304, | |
| "reference_delta_l2": 0.9657699046880037, | |
| "retained_delta_l2": 0.9163237914003645, | |
| "delta_retention_norm_ratio": 0.9488013521154264, | |
| "delta_cosine": 0.7723764319730205, | |
| "relative_weight_rmse": 0.013146854028355406 | |
| }, | |
| { | |
| "index": 124, | |
| "total": 252, | |
| "projection": "layers.24.self_attn.o_proj", | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "seconds": 0.1284, | |
| "parameters": 16777216, | |
| "reference_delta_l2": 1.9180386974565615, | |
| "retained_delta_l2": 1.778889888553453, | |
| "delta_retention_norm_ratio": 0.9274525539616961, | |
| "delta_cosine": 0.7461192695903845, | |
| "relative_weight_rmse": 0.012580078475031757 | |
| }, | |
| { | |
| "index": 125, | |
| "total": 252, | |
| "projection": "layers.24.self_attn.q_proj", | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "seconds": 0.1279, | |
| "parameters": 16777216, | |
| "reference_delta_l2": 3.0223934430575605, | |
| "retained_delta_l2": 3.019102590965159, | |
| "delta_retention_norm_ratio": 0.998911176802623, | |
| "delta_cosine": 0.8488570729763014, | |
| "relative_weight_rmse": 0.015556414370022116 | |
| }, | |
| { | |
| "index": 126, | |
| "total": 252, | |
| "projection": "layers.24.self_attn.v_proj", | |
| "shape": [ | |
| 1024, | |
| 4096 | |
| ], | |
| "seconds": 0.0294, | |
| "parameters": 4194304, | |
| "reference_delta_l2": 1.0890605670331526, | |
| "retained_delta_l2": 1.0379482359911925, | |
| "delta_retention_norm_ratio": 0.9530675036915516, | |
| "delta_cosine": 0.7793129266050705, | |
| "relative_weight_rmse": 0.013058435495478861 | |
| }, | |
| { | |
| "index": 127, | |
| "total": 252, | |
| "projection": "layers.25.mlp.down_proj", | |
| "shape": [ | |
| 4096, | |
| 12288 | |
| ], | |
| "seconds": 0.4228, | |
| "parameters": 50331648, | |
| "reference_delta_l2": 3.63616379677842, | |
| "retained_delta_l2": 3.4026856581495566, | |
| "delta_retention_norm_ratio": 0.9357899831587012, | |
| "delta_cosine": 0.75136838655561, | |
| "relative_weight_rmse": 0.013261149732231414 | |
| }, | |
| { | |
| "index": 128, | |
| "total": 252, | |
| "projection": "layers.25.mlp.gate_proj", | |
| "shape": [ | |
| 12288, | |
| 4096 | |
| ], | |
| "seconds": 0.3792, | |
| "parameters": 50331648, | |
| "reference_delta_l2": 5.588137120575692, | |
| "retained_delta_l2": 5.600545330822972, | |
| "delta_retention_norm_ratio": 1.0022204555793008, | |
| "delta_cosine": 0.8402136399093728, | |
| "relative_weight_rmse": 0.01693661749294787 | |
| }, | |
| { | |
| "index": 129, | |
| "total": 252, | |
| "projection": "layers.25.mlp.up_proj", | |
| "shape": [ | |
| 12288, | |
| 4096 | |
| ], | |
| "seconds": 0.3774, | |
| "parameters": 50331648, | |
| "reference_delta_l2": 3.744055871537401, | |
| "retained_delta_l2": 3.4942314485826187, | |
| "delta_retention_norm_ratio": 0.9332743870480228, | |
| "delta_cosine": 0.7482273397250873, | |
| "relative_weight_rmse": 0.013618174422273666 | |
| }, | |
| { | |
| "index": 130, | |
| "total": 252, | |
| "projection": "layers.25.self_attn.k_proj", | |
| "shape": [ | |
| 1024, | |
| 4096 | |
| ], | |
| "seconds": 0.0307, | |
| "parameters": 4194304, | |
| "reference_delta_l2": 1.0283780121709922, | |
| "retained_delta_l2": 0.9818159558303128, | |
| "delta_retention_norm_ratio": 0.9547228200237546, | |
| "delta_cosine": 0.779950811230599, | |
| "relative_weight_rmse": 0.013540249177058672 | |
| }, | |
| { | |
| "index": 131, | |
| "total": 252, | |
| "projection": "layers.25.self_attn.o_proj", | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "seconds": 0.1276, | |
| "parameters": 16777216, | |
| "reference_delta_l2": 1.9550480162383896, | |
| "retained_delta_l2": 1.8079483667265448, | |
| "delta_retention_norm_ratio": 0.9247590604987431, | |
| "delta_cosine": 0.7401215782827391, | |
| "relative_weight_rmse": 0.013194625829935585 | |
| }, | |
| { | |
| "index": 132, | |
| "total": 252, | |
| "projection": "layers.25.self_attn.q_proj", | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "seconds": 0.1269, | |
| "parameters": 16777216, | |
| "reference_delta_l2": 2.4471720032329207, | |
| "retained_delta_l2": 2.399520063203793, | |
| "delta_retention_norm_ratio": 0.9805277520475981, | |
| "delta_cosine": 0.8149741021838156, | |
| "relative_weight_rmse": 0.014585611942000051 | |
| }, | |
| { | |
| "index": 133, | |
| "total": 252, | |
| "projection": "layers.25.self_attn.v_proj", | |
| "shape": [ | |
| 1024, | |
| 4096 | |
| ], | |
| "seconds": 0.0289, | |
| "parameters": 4194304, | |
| "reference_delta_l2": 1.1872069387379602, | |
| "retained_delta_l2": 1.1437759901729765, | |
| "delta_retention_norm_ratio": 0.9634175415019455, | |
| "delta_cosine": 0.7871225418419032, | |
| "relative_weight_rmse": 0.013451197170588274 | |
| }, | |
| { | |
| "index": 134, | |
| "total": 252, | |
| "projection": "layers.26.mlp.down_proj", | |
| "shape": [ | |
| 4096, | |
| 12288 | |
| ], | |
| "seconds": 0.4216, | |
| "parameters": 50331648, | |
| "reference_delta_l2": 3.2948821444778953, | |
| "retained_delta_l2": 3.0158551550364296, | |
| "delta_retention_norm_ratio": 0.9153150318565705, | |
| "delta_cosine": 0.728234076979134, | |
| "relative_weight_rmse": 0.012335675013045194 | |
| }, | |
| { | |
| "index": 135, | |
| "total": 252, | |
| "projection": "layers.26.mlp.gate_proj", | |
| "shape": [ | |
| 12288, | |
| 4096 | |
| ], | |
| "seconds": 0.3819, | |
| "parameters": 50331648, | |
| "reference_delta_l2": 5.257820107805619, | |
| "retained_delta_l2": 5.22001821197644, | |
| "delta_retention_norm_ratio": 0.9928103481948612, | |
| "delta_cosine": 0.8220782001203848, | |
| "relative_weight_rmse": 0.01665213701027111 | |
| }, | |
| { | |
| "index": 136, | |
| "total": 252, | |
| "projection": "layers.26.mlp.up_proj", | |
| "shape": [ | |
| 12288, | |
| 4096 | |
| ], | |
| "seconds": 0.3807, | |
| "parameters": 50331648, | |
| "reference_delta_l2": 3.5989018785770184, | |
| "retained_delta_l2": 3.3268782382218878, | |
| "delta_retention_norm_ratio": 0.9244148216503509, | |
| "delta_cosine": 0.738176113660347, | |
| "relative_weight_rmse": 0.01316718412135077 | |
| }, | |
| { | |
| "index": 137, | |
| "total": 252, | |
| "projection": "layers.26.self_attn.k_proj", | |
| "shape": [ | |
| 1024, | |
| 4096 | |
| ], | |
| "seconds": 0.0307, | |
| "parameters": 4194304, | |
| "reference_delta_l2": 1.1017757878657917, | |
| "retained_delta_l2": 1.0633556595201925, | |
| "delta_retention_norm_ratio": 0.9651289048382327, | |
| "delta_cosine": 0.7912559274637104, | |
| "relative_weight_rmse": 0.014435964750473713 | |
| }, | |
| { | |
| "index": 138, | |
| "total": 252, | |
| "projection": "layers.26.self_attn.o_proj", | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "seconds": 0.1282, | |
| "parameters": 16777216, | |
| "reference_delta_l2": 2.14452140481544, | |
| "retained_delta_l2": 2.0199786558489006, | |
| "delta_retention_norm_ratio": 0.9419251546350231, | |
| "delta_cosine": 0.7603254744735352, | |
| "relative_weight_rmse": 0.013806357027894066 | |
| }, | |
| { | |
| "index": 139, | |
| "total": 252, | |
| "projection": "layers.26.self_attn.q_proj", | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "seconds": 0.1278, | |
| "parameters": 16777216, | |
| "reference_delta_l2": 2.7102624259983674, | |
| "retained_delta_l2": 2.6858010439526647, | |
| "delta_retention_norm_ratio": 0.9909745337532428, | |
| "delta_cosine": 0.8318261324732998, | |
| "relative_weight_rmse": 0.015352093328554573 | |
| }, | |
| { | |
| "index": 140, | |
| "total": 252, | |
| "projection": "layers.26.self_attn.v_proj", | |
| "shape": [ | |
| 1024, | |
| 4096 | |
| ], | |
| "seconds": 0.0292, | |
| "parameters": 4194304, | |
| "reference_delta_l2": 1.3006259143024874, | |
| "retained_delta_l2": 1.2651158277502554, | |
| "delta_retention_norm_ratio": 0.9726976941165473, | |
| "delta_cosine": 0.8067860578136268, | |
| "relative_weight_rmse": 0.014257367480946508 | |
| }, | |
| { | |
| "index": 141, | |
| "total": 252, | |
| "projection": "layers.27.mlp.down_proj", | |
| "shape": [ | |
| 4096, | |
| 12288 | |
| ], | |
| "seconds": 0.4172, | |
| "parameters": 50331648, | |
| "reference_delta_l2": 3.1210244717525866, | |
| "retained_delta_l2": 2.7939265449511432, | |
| "delta_retention_norm_ratio": 0.8951953341725115, | |
| "delta_cosine": 0.707603604841936, | |
| "relative_weight_rmse": 0.011908033331055366 | |
| }, | |
| { | |
| "index": 142, | |
| "total": 252, | |
| "projection": "layers.27.mlp.gate_proj", | |
| "shape": [ | |
| 12288, | |
| 4096 | |
| ], | |
| "seconds": 0.376, | |
| "parameters": 50331648, | |
| "reference_delta_l2": 5.223471568922569, | |
| "retained_delta_l2": 5.190236864464552, | |
| "delta_retention_norm_ratio": 0.9936374298166475, | |
| "delta_cosine": 0.8248092087450305, | |
| "relative_weight_rmse": 0.016444172947481323 | |
| }, | |
| { | |
| "index": 143, | |
| "total": 252, | |
| "projection": "layers.27.mlp.up_proj", | |
| "shape": [ | |
| 12288, | |
| 4096 | |
| ], | |
| "seconds": 0.3804, | |
| "parameters": 50331648, | |
| "reference_delta_l2": 3.1443452802142606, | |
| "retained_delta_l2": 2.7614216127494564, | |
| "delta_retention_norm_ratio": 0.8782183146760806, | |
| "delta_cosine": 0.6899573747353942, | |
| "relative_weight_rmse": 0.01218497507994894 | |
| }, | |
| { | |
| "index": 144, | |
| "total": 252, | |
| "projection": "layers.27.self_attn.k_proj", | |
| "shape": [ | |
| 1024, | |
| 4096 | |
| ], | |
| "seconds": 0.0301, | |
| "parameters": 4194304, | |
| "reference_delta_l2": 0.9579004907051812, | |
| "retained_delta_l2": 0.9016451638898682, | |
| "delta_retention_norm_ratio": 0.9412722643310273, | |
| "delta_cosine": 0.7665693409053349, | |
| "relative_weight_rmse": 0.013499582772046447 | |
| }, | |
| { | |
| "index": 145, | |
| "total": 252, | |
| "projection": "layers.27.self_attn.o_proj", | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "seconds": 0.1269, | |
| "parameters": 16777216, | |
| "reference_delta_l2": 1.8332854378342283, | |
| "retained_delta_l2": 1.6681613288420845, | |
| "delta_retention_norm_ratio": 0.9099299511224967, | |
| "delta_cosine": 0.7238126461520346, | |
| "relative_weight_rmse": 0.012739701931374197 | |
| }, | |
| { | |
| "index": 146, | |
| "total": 252, | |
| "projection": "layers.27.self_attn.q_proj", | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "seconds": 0.1261, | |
| "parameters": 16777216, | |
| "reference_delta_l2": 2.5066116371408538, | |
| "retained_delta_l2": 2.4520825621119036, | |
| "delta_retention_norm_ratio": 0.978245902069158, | |
| "delta_cosine": 0.813475660318453, | |
| "relative_weight_rmse": 0.01490029214231884 | |
| }, | |
| { | |
| "index": 147, | |
| "total": 252, | |
| "projection": "layers.27.self_attn.v_proj", | |
| "shape": [ | |
| 1024, | |
| 4096 | |
| ], | |
| "seconds": 0.0288, | |
| "parameters": 4194304, | |
| "reference_delta_l2": 1.305228018543387, | |
| "retained_delta_l2": 1.2820434572650723, | |
| "delta_retention_norm_ratio": 0.9822371563060772, | |
| "delta_cosine": 0.8141868320028673, | |
| "relative_weight_rmse": 0.01431323852325054 | |
| }, | |
| { | |
| "index": 148, | |
| "total": 252, | |
| "projection": "layers.28.mlp.down_proj", | |
| "shape": [ | |
| 4096, | |
| 12288 | |
| ], | |
| "seconds": 0.4181, | |
| "parameters": 50331648, | |
| "reference_delta_l2": 3.7609601109774404, | |
| "retained_delta_l2": 3.5458474311195634, | |
| "delta_retention_norm_ratio": 0.9428037858657397, | |
| "delta_cosine": 0.7609299525782532, | |
| "relative_weight_rmse": 0.013099285885407297 | |
| }, | |
| { | |
| "index": 149, | |
| "total": 252, | |
| "projection": "layers.28.mlp.gate_proj", | |
| "shape": [ | |
| 12288, | |
| 4096 | |
| ], | |
| "seconds": 0.3767, | |
| "parameters": 50331648, | |
| "reference_delta_l2": 6.409505864458649, | |
| "retained_delta_l2": 6.522239218789574, | |
| "delta_retention_norm_ratio": 1.0175884626233112, | |
| "delta_cosine": 0.8665051334261239, | |
| "relative_weight_rmse": 0.017929193401937306 | |
| }, | |
| { | |
| "index": 150, | |
| "total": 252, | |
| "projection": "layers.28.mlp.up_proj", | |
| "shape": [ | |
| 12288, | |
| 4096 | |
| ], | |
| "seconds": 0.3797, | |
| "parameters": 50331648, | |
| "reference_delta_l2": 3.240790766573222, | |
| "retained_delta_l2": 2.8726819439189053, | |
| "delta_retention_norm_ratio": 0.8864138881006652, | |
| "delta_cosine": 0.6976513684563926, | |
| "relative_weight_rmse": 0.012317570764878012 | |
| }, | |
| { | |
| "index": 151, | |
| "total": 252, | |
| "projection": "layers.28.self_attn.k_proj", | |
| "shape": [ | |
| 1024, | |
| 4096 | |
| ], | |
| "seconds": 0.0301, | |
| "parameters": 4194304, | |
| "reference_delta_l2": 1.0315480582853214, | |
| "retained_delta_l2": 0.9815406033974483, | |
| "delta_retention_norm_ratio": 0.95152193396496, | |
| "delta_cosine": 0.7709704314678171, | |
| "relative_weight_rmse": 0.014130402020247729 | |
| }, | |
| { | |
| "index": 152, | |
| "total": 252, | |
| "projection": "layers.28.self_attn.o_proj", | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "seconds": 0.1271, | |
| "parameters": 16777216, | |
| "reference_delta_l2": 1.866301512748793, | |
| "retained_delta_l2": 1.714424115865921, | |
| "delta_retention_norm_ratio": 0.9186211896387639, | |
| "delta_cosine": 0.7336246512815289, | |
| "relative_weight_rmse": 0.012828730649353547 | |
| }, | |
| { | |
| "index": 153, | |
| "total": 252, | |
| "projection": "layers.28.self_attn.q_proj", | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "seconds": 0.1276, | |
| "parameters": 16777216, | |
| "reference_delta_l2": 2.1056798548733475, | |
| "retained_delta_l2": 2.0057791667171596, | |
| "delta_retention_norm_ratio": 0.9525565636556862, | |
| "delta_cosine": 0.776883814777253, | |
| "relative_weight_rmse": 0.013245033292334856 | |
| }, | |
| { | |
| "index": 154, | |
| "total": 252, | |
| "projection": "layers.28.self_attn.v_proj", | |
| "shape": [ | |
| 1024, | |
| 4096 | |
| ], | |
| "seconds": 0.0291, | |
| "parameters": 4194304, | |
| "reference_delta_l2": 1.0578716443620564, | |
| "retained_delta_l2": 0.9885759113310327, | |
| "delta_retention_norm_ratio": 0.9344951408799579, | |
| "delta_cosine": 0.7520981682275366, | |
| "relative_weight_rmse": 0.012589562464635296 | |
| }, | |
| { | |
| "index": 155, | |
| "total": 252, | |
| "projection": "layers.29.mlp.down_proj", | |
| "shape": [ | |
| 4096, | |
| 12288 | |
| ], | |
| "seconds": 0.4172, | |
| "parameters": 50331648, | |
| "reference_delta_l2": 3.7400414679280893, | |
| "retained_delta_l2": 3.5093891045814924, | |
| "delta_retention_norm_ratio": 0.9383289288836753, | |
| "delta_cosine": 0.7558183147959916, | |
| "relative_weight_rmse": 0.012999432967736499 | |
| }, | |
| { | |
| "index": 156, | |
| "total": 252, | |
| "projection": "layers.29.mlp.gate_proj", | |
| "shape": [ | |
| 12288, | |
| 4096 | |
| ], | |
| "seconds": 0.3765, | |
| "parameters": 50331648, | |
| "reference_delta_l2": 6.119080985772878, | |
| "retained_delta_l2": 6.202541880165826, | |
| "delta_retention_norm_ratio": 1.0136394492223584, | |
| "delta_cosine": 0.8565729728321285, | |
| "relative_weight_rmse": 0.017705159618673742 | |
| }, | |
| { | |
| "index": 157, | |
| "total": 252, | |
| "projection": "layers.29.mlp.up_proj", | |
| "shape": [ | |
| 12288, | |
| 4096 | |
| ], | |
| "seconds": 0.377, | |
| "parameters": 50331648, | |
| "reference_delta_l2": 3.5792385362284014, | |
| "retained_delta_l2": 3.278999355649104, | |
| "delta_retention_norm_ratio": 0.9161164651251011, | |
| "delta_cosine": 0.7293019135804754, | |
| "relative_weight_rmse": 0.012895063583214437 | |
| }, | |
| { | |
| "index": 158, | |
| "total": 252, | |
| "projection": "layers.29.self_attn.k_proj", | |
| "shape": [ | |
| 1024, | |
| 4096 | |
| ], | |
| "seconds": 0.0335, | |
| "parameters": 4194304, | |
| "reference_delta_l2": 0.9611100218066246, | |
| "retained_delta_l2": 0.9109171148597183, | |
| "delta_retention_norm_ratio": 0.9477761069928733, | |
| "delta_cosine": 0.7815760941041824, | |
| "relative_weight_rmse": 0.013125099298455387 | |
| }, | |
| { | |
| "index": 159, | |
| "total": 252, | |
| "projection": "layers.29.self_attn.o_proj", | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "seconds": 0.1298, | |
| "parameters": 16777216, | |
| "reference_delta_l2": 1.9448868287407715, | |
| "retained_delta_l2": 1.7931075348352579, | |
| "delta_retention_norm_ratio": 0.9219598324886678, | |
| "delta_cosine": 0.7375264672159623, | |
| "relative_weight_rmse": 0.012926364794946505 | |
| }, | |
| { | |
| "index": 160, | |
| "total": 252, | |
| "projection": "layers.29.self_attn.q_proj", | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "seconds": 0.1282, | |
| "parameters": 16777216, | |
| "reference_delta_l2": 2.9410678323592614, | |
| "retained_delta_l2": 2.927741570929101, | |
| "delta_retention_norm_ratio": 0.995468903748653, | |
| "delta_cosine": 0.8476177653552707, | |
| "relative_weight_rmse": 0.015271058137973317 | |
| }, | |
| { | |
| "index": 161, | |
| "total": 252, | |
| "projection": "layers.29.self_attn.v_proj", | |
| "shape": [ | |
| 1024, | |
| 4096 | |
| ], | |
| "seconds": 0.0291, | |
| "parameters": 4194304, | |
| "reference_delta_l2": 1.0694664804434102, | |
| "retained_delta_l2": 1.010059414365901, | |
| "delta_retention_norm_ratio": 0.9444516801939613, | |
| "delta_cosine": 0.7675995472724292, | |
| "relative_weight_rmse": 0.01295360202158742 | |
| }, | |
| { | |
| "index": 162, | |
| "total": 252, | |
| "projection": "layers.3.mlp.down_proj", | |
| "shape": [ | |
| 4096, | |
| 12288 | |
| ], | |
| "seconds": 0.4203, | |
| "parameters": 50331648, | |
| "reference_delta_l2": 2.0612252225954784, | |
| "retained_delta_l2": 1.8231090160941608, | |
| "delta_retention_norm_ratio": 0.8844783171236942, | |
| "delta_cosine": 0.7151755493808815, | |
| "relative_weight_rmse": 0.01132277695851018 | |
| }, | |
| { | |
| "index": 163, | |
| "total": 252, | |
| "projection": "layers.3.mlp.gate_proj", | |
| "shape": [ | |
| 12288, | |
| 4096 | |
| ], | |
| "seconds": 0.3757, | |
| "parameters": 50331648, | |
| "reference_delta_l2": 2.9220544512009474, | |
| "retained_delta_l2": 2.7004006597997616, | |
| "delta_retention_norm_ratio": 0.9241445376522374, | |
| "delta_cosine": 0.7457059707004465, | |
| "relative_weight_rmse": 0.012247179531549551 | |
| }, | |
| { | |
| "index": 164, | |
| "total": 252, | |
| "projection": "layers.3.mlp.up_proj", | |
| "shape": [ | |
| 12288, | |
| 4096 | |
| ], | |
| "seconds": 0.3801, | |
| "parameters": 50331648, | |
| "reference_delta_l2": 3.1727358381363153, | |
| "retained_delta_l2": 3.106383395466854, | |
| "delta_retention_norm_ratio": 0.9790866791139986, | |
| "delta_cosine": 0.8418774960196438, | |
| "relative_weight_rmse": 0.014668988845837653 | |
| }, | |
| { | |
| "index": 165, | |
| "total": 252, | |
| "projection": "layers.3.self_attn.k_proj", | |
| "shape": [ | |
| 1024, | |
| 4096 | |
| ], | |
| "seconds": 0.0307, | |
| "parameters": 4194304, | |
| "reference_delta_l2": 0.5945609896406548, | |
| "retained_delta_l2": 0.4614519112115394, | |
| "delta_retention_norm_ratio": 0.7761220787297786, | |
| "delta_cosine": 0.5970850492537589, | |
| "relative_weight_rmse": 0.009637125549736558 | |
| }, | |
| { | |
| "index": 166, | |
| "total": 252, | |
| "projection": "layers.3.self_attn.o_proj", | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "seconds": 0.1306, | |
| "parameters": 16777216, | |
| "reference_delta_l2": 1.8551680918690623, | |
| "retained_delta_l2": 1.702378465435281, | |
| "delta_retention_norm_ratio": 0.9176410875632044, | |
| "delta_cosine": 0.7310633062118554, | |
| "relative_weight_rmse": 0.012836174509958146 | |
| }, | |
| { | |
| "index": 167, | |
| "total": 252, | |
| "projection": "layers.3.self_attn.q_proj", | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "seconds": 0.1283, | |
| "parameters": 16777216, | |
| "reference_delta_l2": 1.50916759836078, | |
| "retained_delta_l2": 1.3066576875077118, | |
| "delta_retention_norm_ratio": 0.8658135046942239, | |
| "delta_cosine": 0.6791651347451496, | |
| "relative_weight_rmse": 0.011729254761278396 | |
| }, | |
| { | |
| "index": 168, | |
| "total": 252, | |
| "projection": "layers.3.self_attn.v_proj", | |
| "shape": [ | |
| 1024, | |
| 4096 | |
| ], | |
| "seconds": 0.0304, | |
| "parameters": 4194304, | |
| "reference_delta_l2": 0.9493558174419165, | |
| "retained_delta_l2": 0.8560802908478509, | |
| "delta_retention_norm_ratio": 0.9017486121848381, | |
| "delta_cosine": 0.7160920598769267, | |
| "relative_weight_rmse": 0.012411782534285076 | |
| }, | |
| { | |
| "index": 169, | |
| "total": 252, | |
| "projection": "layers.30.mlp.down_proj", | |
| "shape": [ | |
| 4096, | |
| 12288 | |
| ], | |
| "seconds": 0.426, | |
| "parameters": 50331648, | |
| "reference_delta_l2": 3.891216371645432, | |
| "retained_delta_l2": 3.691144031245628, | |
| "delta_retention_norm_ratio": 0.9485835992422077, | |
| "delta_cosine": 0.7667413673469646, | |
| "relative_weight_rmse": 0.01314762686300376 | |
| }, | |
| { | |
| "index": 170, | |
| "total": 252, | |
| "projection": "layers.30.mlp.gate_proj", | |
| "shape": [ | |
| 12288, | |
| 4096 | |
| ], | |
| "seconds": 0.3781, | |
| "parameters": 50331648, | |
| "reference_delta_l2": 5.390923699257254, | |
| "retained_delta_l2": 5.38683376050761, | |
| "delta_retention_norm_ratio": 0.9992413287633423, | |
| "delta_cosine": 0.8303979683713872, | |
| "relative_weight_rmse": 0.016918480433160996 | |
| }, | |
| { | |
| "index": 171, | |
| "total": 252, | |
| "projection": "layers.30.mlp.up_proj", | |
| "shape": [ | |
| 12288, | |
| 4096 | |
| ], | |
| "seconds": 0.3779, | |
| "parameters": 50331648, | |
| "reference_delta_l2": 3.803511864260935, | |
| "retained_delta_l2": 3.5221099707322403, | |
| "delta_retention_norm_ratio": 0.9260152449706176, | |
| "delta_cosine": 0.7392997590801647, | |
| "relative_weight_rmse": 0.013392195073316317 | |
| }, | |
| { | |
| "index": 172, | |
| "total": 252, | |
| "projection": "layers.30.self_attn.k_proj", | |
| "shape": [ | |
| 1024, | |
| 4096 | |
| ], | |
| "seconds": 0.0304, | |
| "parameters": 4194304, | |
| "reference_delta_l2": 1.493770964022566, | |
| "retained_delta_l2": 1.5055196729336286, | |
| "delta_retention_norm_ratio": 1.0078651340761267, | |
| "delta_cosine": 0.8526237968690384, | |
| "relative_weight_rmse": 0.01659810135995429 | |
| }, | |
| { | |
| "index": 173, | |
| "total": 252, | |
| "projection": "layers.30.self_attn.o_proj", | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "seconds": 0.13, | |
| "parameters": 16777216, | |
| "reference_delta_l2": 1.869066986144252, | |
| "retained_delta_l2": 1.6929682622874866, | |
| "delta_retention_norm_ratio": 0.9057825507794967, | |
| "delta_cosine": 0.7176126199142076, | |
| "relative_weight_rmse": 0.012864923697526695 | |
| }, | |
| { | |
| "index": 174, | |
| "total": 252, | |
| "projection": "layers.30.self_attn.q_proj", | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "seconds": 0.1274, | |
| "parameters": 16777216, | |
| "reference_delta_l2": 3.0153665146577913, | |
| "retained_delta_l2": 3.026151166028698, | |
| "delta_retention_norm_ratio": 1.0035765640158443, | |
| "delta_cosine": 0.8548589463251729, | |
| "relative_weight_rmse": 0.016286740512417037 | |
| }, | |
| { | |
| "index": 175, | |
| "total": 252, | |
| "projection": "layers.30.self_attn.v_proj", | |
| "shape": [ | |
| 1024, | |
| 4096 | |
| ], | |
| "seconds": 0.0293, | |
| "parameters": 4194304, | |
| "reference_delta_l2": 1.3107439157647822, | |
| "retained_delta_l2": 1.2728059144749695, | |
| "delta_retention_norm_ratio": 0.9710561301612628, | |
| "delta_cosine": 0.8016661063381451, | |
| "relative_weight_rmse": 0.013927134335966412 | |
| }, | |
| { | |
| "index": 176, | |
| "total": 252, | |
| "projection": "layers.31.mlp.down_proj", | |
| "shape": [ | |
| 4096, | |
| 12288 | |
| ], | |
| "seconds": 0.4218, | |
| "parameters": 50331648, | |
| "reference_delta_l2": 4.112663100444478, | |
| "retained_delta_l2": 3.947313056520835, | |
| "delta_retention_norm_ratio": 0.9597948969110129, | |
| "delta_cosine": 0.7806269001044362, | |
| "relative_weight_rmse": 0.013383008011024313 | |
| }, | |
| { | |
| "index": 177, | |
| "total": 252, | |
| "projection": "layers.31.mlp.gate_proj", | |
| "shape": [ | |
| 12288, | |
| 4096 | |
| ], | |
| "seconds": 0.3774, | |
| "parameters": 50331648, | |
| "reference_delta_l2": 4.207933372829771, | |
| "retained_delta_l2": 4.058604717995024, | |
| "delta_retention_norm_ratio": 0.9645125904799376, | |
| "delta_cosine": 0.785362693443029, | |
| "relative_weight_rmse": 0.014746817633539662 | |
| }, | |
| { | |
| "index": 178, | |
| "total": 252, | |
| "projection": "layers.31.mlp.up_proj", | |
| "shape": [ | |
| 12288, | |
| 4096 | |
| ], | |
| "seconds": 0.3773, | |
| "parameters": 50331648, | |
| "reference_delta_l2": 3.744293271609304, | |
| "retained_delta_l2": 3.4567529672151363, | |
| "delta_retention_norm_ratio": 0.9232057204027231, | |
| "delta_cosine": 0.7366416220496778, | |
| "relative_weight_rmse": 0.013122450308699104 | |
| }, | |
| { | |
| "index": 179, | |
| "total": 252, | |
| "projection": "layers.31.self_attn.k_proj", | |
| "shape": [ | |
| 1024, | |
| 4096 | |
| ], | |
| "seconds": 0.0304, | |
| "parameters": 4194304, | |
| "reference_delta_l2": 1.1306844138330658, | |
| "retained_delta_l2": 1.1067167777558367, | |
| "delta_retention_norm_ratio": 0.9788025413775909, | |
| "delta_cosine": 0.8128010662192252, | |
| "relative_weight_rmse": 0.014431528691286433 | |
| }, | |
| { | |
| "index": 180, | |
| "total": 252, | |
| "projection": "layers.31.self_attn.o_proj", | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "seconds": 0.1291, | |
| "parameters": 16777216, | |
| "reference_delta_l2": 1.9262812779313172, | |
| "retained_delta_l2": 1.7774918803748536, | |
| "delta_retention_norm_ratio": 0.9227582185109268, | |
| "delta_cosine": 0.7387240864420156, | |
| "relative_weight_rmse": 0.01253640903866733 | |
| }, | |
| { | |
| "index": 181, | |
| "total": 252, | |
| "projection": "layers.31.self_attn.q_proj", | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "seconds": 0.1277, | |
| "parameters": 16777216, | |
| "reference_delta_l2": 3.091232754030015, | |
| "retained_delta_l2": 3.114182056358259, | |
| "delta_retention_norm_ratio": 1.0074239968822551, | |
| "delta_cosine": 0.859119072636568, | |
| "relative_weight_rmse": 0.016349614371943485 | |
| }, | |
| { | |
| "index": 182, | |
| "total": 252, | |
| "projection": "layers.31.self_attn.v_proj", | |
| "shape": [ | |
| 1024, | |
| 4096 | |
| ], | |
| "seconds": 0.0293, | |
| "parameters": 4194304, | |
| "reference_delta_l2": 0.9129372319850592, | |
| "retained_delta_l2": 0.80527012589888, | |
| "delta_retention_norm_ratio": 0.8820651603264427, | |
| "delta_cosine": 0.699342117861831, | |
| "relative_weight_rmse": 0.011244243849305599 | |
| }, | |
| { | |
| "index": 183, | |
| "total": 252, | |
| "projection": "layers.32.mlp.down_proj", | |
| "shape": [ | |
| 4096, | |
| 12288 | |
| ], | |
| "seconds": 0.4188, | |
| "parameters": 50331648, | |
| "reference_delta_l2": 3.0531319642760333, | |
| "retained_delta_l2": 2.700968335821406, | |
| "delta_retention_norm_ratio": 0.884654960029501, | |
| "delta_cosine": 0.6983479871907332, | |
| "relative_weight_rmse": 0.011288085442385564 | |
| }, | |
| { | |
| "index": 184, | |
| "total": 252, | |
| "projection": "layers.32.mlp.gate_proj", | |
| "shape": [ | |
| 12288, | |
| 4096 | |
| ], | |
| "seconds": 0.3821, | |
| "parameters": 50331648, | |
| "reference_delta_l2": 5.724801736222787, | |
| "retained_delta_l2": 5.7750467166699035, | |
| "delta_retention_norm_ratio": 1.0087767197472008, | |
| "delta_cosine": 0.8479988489236641, | |
| "relative_weight_rmse": 0.017468866184904154 | |
| }, | |
| { | |
| "index": 185, | |
| "total": 252, | |
| "projection": "layers.32.mlp.up_proj", | |
| "shape": [ | |
| 12288, | |
| 4096 | |
| ], | |
| "seconds": 0.3873, | |
| "parameters": 50331648, | |
| "reference_delta_l2": 4.175474368127258, | |
| "retained_delta_l2": 3.9589358467876394, | |
| "delta_retention_norm_ratio": 0.9481403782543784, | |
| "delta_cosine": 0.7671152769314903, | |
| "relative_weight_rmse": 0.013828873112894892 | |
| }, | |
| { | |
| "index": 186, | |
| "total": 252, | |
| "projection": "layers.32.self_attn.k_proj", | |
| "shape": [ | |
| 1024, | |
| 4096 | |
| ], | |
| "seconds": 0.0343, | |
| "parameters": 4194304, | |
| "reference_delta_l2": 1.0750931486612259, | |
| "retained_delta_l2": 1.040542228910182, | |
| "delta_retention_norm_ratio": 0.9678623942547965, | |
| "delta_cosine": 0.7986368381511688, | |
| "relative_weight_rmse": 0.014081564546770174 | |
| }, | |
| { | |
| "index": 187, | |
| "total": 252, | |
| "projection": "layers.32.self_attn.o_proj", | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "seconds": 0.1296, | |
| "parameters": 16777216, | |
| "reference_delta_l2": 1.9535348950549298, | |
| "retained_delta_l2": 1.7957168772677157, | |
| "delta_retention_norm_ratio": 0.9192141291221846, | |
| "delta_cosine": 0.7328067682184131, | |
| "relative_weight_rmse": 0.012910728503809407 | |
| }, | |
| { | |
| "index": 188, | |
| "total": 252, | |
| "projection": "layers.32.self_attn.q_proj", | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "seconds": 0.127, | |
| "parameters": 16777216, | |
| "reference_delta_l2": 3.498986324699088, | |
| "retained_delta_l2": 3.558383258594726, | |
| "delta_retention_norm_ratio": 1.0169754690026536, | |
| "delta_cosine": 0.8835599014845256, | |
| "relative_weight_rmse": 0.017005628509784897 | |
| }, | |
| { | |
| "index": 189, | |
| "total": 252, | |
| "projection": "layers.32.self_attn.v_proj", | |
| "shape": [ | |
| 1024, | |
| 4096 | |
| ], | |
| "seconds": 0.0293, | |
| "parameters": 4194304, | |
| "reference_delta_l2": 1.1058993601101823, | |
| "retained_delta_l2": 1.0362872940572145, | |
| "delta_retention_norm_ratio": 0.937053887031789, | |
| "delta_cosine": 0.7581904018210112, | |
| "relative_weight_rmse": 0.01261230357564126 | |
| }, | |
| { | |
| "index": 190, | |
| "total": 252, | |
| "projection": "layers.33.mlp.down_proj", | |
| "shape": [ | |
| 4096, | |
| 12288 | |
| ], | |
| "seconds": 0.42, | |
| "parameters": 50331648, | |
| "reference_delta_l2": 4.680454662657774, | |
| "retained_delta_l2": 4.59402019908065, | |
| "delta_retention_norm_ratio": 0.9815328916084314, | |
| "delta_cosine": 0.809854469246563, | |
| "relative_weight_rmse": 0.01440706327110179 | |
| }, | |
| { | |
| "index": 191, | |
| "total": 252, | |
| "projection": "layers.33.mlp.gate_proj", | |
| "shape": [ | |
| 12288, | |
| 4096 | |
| ], | |
| "seconds": 0.3783, | |
| "parameters": 50331648, | |
| "reference_delta_l2": 3.9928551759715663, | |
| "retained_delta_l2": 3.8067651307265034, | |
| "delta_retention_norm_ratio": 0.9533942412024042, | |
| "delta_cosine": 0.7702773788126092, | |
| "relative_weight_rmse": 0.014707472638971035 | |
| }, | |
| { | |
| "index": 192, | |
| "total": 252, | |
| "projection": "layers.33.mlp.up_proj", | |
| "shape": [ | |
| 12288, | |
| 4096 | |
| ], | |
| "seconds": 0.3806, | |
| "parameters": 50331648, | |
| "reference_delta_l2": 4.544459331194087, | |
| "retained_delta_l2": 4.40694759884874, | |
| "delta_retention_norm_ratio": 0.9697407937174312, | |
| "delta_cosine": 0.795194596650398, | |
| "relative_weight_rmse": 0.014216056005712982 | |
| }, | |
| { | |
| "index": 193, | |
| "total": 252, | |
| "projection": "layers.33.self_attn.k_proj", | |
| "shape": [ | |
| 1024, | |
| 4096 | |
| ], | |
| "seconds": 0.0305, | |
| "parameters": 4194304, | |
| "reference_delta_l2": 1.1713470907351642, | |
| "retained_delta_l2": 1.1620024134850435, | |
| "delta_retention_norm_ratio": 0.9920222815901171, | |
| "delta_cosine": 0.8277024355769212, | |
| "relative_weight_rmse": 0.015548115562863228 | |
| }, | |
| { | |
| "index": 194, | |
| "total": 252, | |
| "projection": "layers.33.self_attn.o_proj", | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "seconds": 0.1278, | |
| "parameters": 16777216, | |
| "reference_delta_l2": 1.9580319646659914, | |
| "retained_delta_l2": 1.810232136079443, | |
| "delta_retention_norm_ratio": 0.9245161308631851, | |
| "delta_cosine": 0.741492899620641, | |
| "relative_weight_rmse": 0.012302995093158103 | |
| }, | |
| { | |
| "index": 195, | |
| "total": 252, | |
| "projection": "layers.33.self_attn.q_proj", | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "seconds": 0.1281, | |
| "parameters": 16777216, | |
| "reference_delta_l2": 1.6943967329802574, | |
| "retained_delta_l2": 1.562015056610022, | |
| "delta_retention_norm_ratio": 0.9218709091008511, | |
| "delta_cosine": 0.7426421898099466, | |
| "relative_weight_rmse": 0.011918602777488662 | |
| }, | |
| { | |
| "index": 196, | |
| "total": 252, | |
| "projection": "layers.33.self_attn.v_proj", | |
| "shape": [ | |
| 1024, | |
| 4096 | |
| ], | |
| "seconds": 0.0293, | |
| "parameters": 4194304, | |
| "reference_delta_l2": 1.2362498105050221, | |
| "retained_delta_l2": 1.1869907840241474, | |
| "delta_retention_norm_ratio": 0.9601544719665098, | |
| "delta_cosine": 0.7867592705647769, | |
| "relative_weight_rmse": 0.012739169630235064 | |
| }, | |
| { | |
| "index": 197, | |
| "total": 252, | |
| "projection": "layers.34.mlp.down_proj", | |
| "shape": [ | |
| 4096, | |
| 12288 | |
| ], | |
| "seconds": 0.42, | |
| "parameters": 50331648, | |
| "reference_delta_l2": 3.248710622949176, | |
| "retained_delta_l2": 3.016860809350351, | |
| "delta_retention_norm_ratio": 0.9286332823979405, | |
| "delta_cosine": 0.7449688814837238, | |
| "relative_weight_rmse": 0.011825474077757182 | |
| }, | |
| { | |
| "index": 198, | |
| "total": 252, | |
| "projection": "layers.34.mlp.gate_proj", | |
| "shape": [ | |
| 12288, | |
| 4096 | |
| ], | |
| "seconds": 0.3793, | |
| "parameters": 50331648, | |
| "reference_delta_l2": 3.5098548391117097, | |
| "retained_delta_l2": 3.3057036996885336, | |
| "delta_retention_norm_ratio": 0.9418348767167695, | |
| "delta_cosine": 0.7568637240827227, | |
| "relative_weight_rmse": 0.013100890845810342 | |
| }, | |
| { | |
| "index": 199, | |
| "total": 252, | |
| "projection": "layers.34.mlp.up_proj", | |
| "shape": [ | |
| 12288, | |
| 4096 | |
| ], | |
| "seconds": 0.3798, | |
| "parameters": 50331648, | |
| "reference_delta_l2": 3.119166573786764, | |
| "retained_delta_l2": 2.7949884408142536, | |
| "delta_retention_norm_ratio": 0.8960689898074445, | |
| "delta_cosine": 0.7095439682266548, | |
| "relative_weight_rmse": 0.011413755648549849 | |
| }, | |
| { | |
| "index": 200, | |
| "total": 252, | |
| "projection": "layers.34.self_attn.k_proj", | |
| "shape": [ | |
| 1024, | |
| 4096 | |
| ], | |
| "seconds": 0.0306, | |
| "parameters": 4194304, | |
| "reference_delta_l2": 0.788059537523479, | |
| "retained_delta_l2": 0.7264836013467776, | |
| "delta_retention_norm_ratio": 0.9218638526091477, | |
| "delta_cosine": 0.7523719001179419, | |
| "relative_weight_rmse": 0.011623414101459404 | |
| }, | |
| { | |
| "index": 201, | |
| "total": 252, | |
| "projection": "layers.34.self_attn.o_proj", | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "seconds": 0.1286, | |
| "parameters": 16777216, | |
| "reference_delta_l2": 2.366467424692969, | |
| "retained_delta_l2": 2.295917305881346, | |
| "delta_retention_norm_ratio": 0.970187580832313, | |
| "delta_cosine": 0.7929682214572963, | |
| "relative_weight_rmse": 0.014046097788924336 | |
| }, | |
| { | |
| "index": 202, | |
| "total": 252, | |
| "projection": "layers.34.self_attn.q_proj", | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "seconds": 0.1293, | |
| "parameters": 16777216, | |
| "reference_delta_l2": 2.790418420753438, | |
| "retained_delta_l2": 2.803650672675589, | |
| "delta_retention_norm_ratio": 1.0047420314544004, | |
| "delta_cosine": 0.8597101316514167, | |
| "relative_weight_rmse": 0.015277924377240636 | |
| }, | |
| { | |
| "index": 203, | |
| "total": 252, | |
| "projection": "layers.34.self_attn.v_proj", | |
| "shape": [ | |
| 1024, | |
| 4096 | |
| ], | |
| "seconds": 0.0293, | |
| "parameters": 4194304, | |
| "reference_delta_l2": 0.9825920197694892, | |
| "retained_delta_l2": 0.8780767742150043, | |
| "delta_retention_norm_ratio": 0.893633121934978, | |
| "delta_cosine": 0.7155778670479893, | |
| "relative_weight_rmse": 0.011416316401979395 | |
| }, | |
| { | |
| "index": 204, | |
| "total": 252, | |
| "projection": "layers.35.mlp.down_proj", | |
| "shape": [ | |
| 4096, | |
| 12288 | |
| ], | |
| "seconds": 0.4207, | |
| "parameters": 50331648, | |
| "reference_delta_l2": 2.331282571631798, | |
| "retained_delta_l2": 2.0846773160858074, | |
| "delta_retention_norm_ratio": 0.8942190626967292, | |
| "delta_cosine": 0.71456817683431, | |
| "relative_weight_rmse": 0.009809271791772265 | |
| }, | |
| { | |
| "index": 205, | |
| "total": 252, | |
| "projection": "layers.35.mlp.gate_proj", | |
| "shape": [ | |
| 12288, | |
| 4096 | |
| ], | |
| "seconds": 0.3807, | |
| "parameters": 50331648, | |
| "reference_delta_l2": 2.6185183399349397, | |
| "retained_delta_l2": 2.2835312544709794, | |
| "delta_retention_norm_ratio": 0.8720699869253986, | |
| "delta_cosine": 0.6867799203989726, | |
| "relative_weight_rmse": 0.010510595372961532 | |
| }, | |
| { | |
| "index": 206, | |
| "total": 252, | |
| "projection": "layers.35.mlp.up_proj", | |
| "shape": [ | |
| 12288, | |
| 4096 | |
| ], | |
| "seconds": 0.3795, | |
| "parameters": 50331648, | |
| "reference_delta_l2": 2.6570990323321224, | |
| "retained_delta_l2": 2.2838320956928437, | |
| "delta_retention_norm_ratio": 0.8595208789370323, | |
| "delta_cosine": 0.674113310648716, | |
| "relative_weight_rmse": 0.010333697702349381 | |
| }, | |
| { | |
| "index": 207, | |
| "total": 252, | |
| "projection": "layers.35.self_attn.k_proj", | |
| "shape": [ | |
| 1024, | |
| 4096 | |
| ], | |
| "seconds": 0.0303, | |
| "parameters": 4194304, | |
| "reference_delta_l2": 0.8324669053364727, | |
| "retained_delta_l2": 0.7624225034713568, | |
| "delta_retention_norm_ratio": 0.9158592354649763, | |
| "delta_cosine": 0.7341526429180774, | |
| "relative_weight_rmse": 0.012225562971949431 | |
| }, | |
| { | |
| "index": 208, | |
| "total": 252, | |
| "projection": "layers.35.self_attn.o_proj", | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "seconds": 0.1284, | |
| "parameters": 16777216, | |
| "reference_delta_l2": 1.9803735222413514, | |
| "retained_delta_l2": 1.879211962319679, | |
| "delta_retention_norm_ratio": 0.9489179395777926, | |
| "delta_cosine": 0.7677733338888255, | |
| "relative_weight_rmse": 0.012993498178701863 | |
| }, | |
| { | |
| "index": 209, | |
| "total": 252, | |
| "projection": "layers.35.self_attn.q_proj", | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "seconds": 0.1291, | |
| "parameters": 16777216, | |
| "reference_delta_l2": 1.9183182575278923, | |
| "retained_delta_l2": 1.8299753931083624, | |
| "delta_retention_norm_ratio": 0.9539477539387151, | |
| "delta_cosine": 0.7741720514765471, | |
| "relative_weight_rmse": 0.013070551367355533 | |
| }, | |
| { | |
| "index": 210, | |
| "total": 252, | |
| "projection": "layers.35.self_attn.v_proj", | |
| "shape": [ | |
| 1024, | |
| 4096 | |
| ], | |
| "seconds": 0.0299, | |
| "parameters": 4194304, | |
| "reference_delta_l2": 0.904757871002193, | |
| "retained_delta_l2": 0.7982918615926153, | |
| "delta_retention_norm_ratio": 0.8823265176000668, | |
| "delta_cosine": 0.6993180987503474, | |
| "relative_weight_rmse": 0.011424754057315271 | |
| }, | |
| { | |
| "index": 211, | |
| "total": 252, | |
| "projection": "layers.4.mlp.down_proj", | |
| "shape": [ | |
| 4096, | |
| 12288 | |
| ], | |
| "seconds": 0.4206, | |
| "parameters": 50331648, | |
| "reference_delta_l2": 2.2235173063590223, | |
| "retained_delta_l2": 1.906815818828342, | |
| "delta_retention_norm_ratio": 0.8575673386373257, | |
| "delta_cosine": 0.6728611526585633, | |
| "relative_weight_rmse": 0.01083163399891114 | |
| }, | |
| { | |
| "index": 212, | |
| "total": 252, | |
| "projection": "layers.4.mlp.gate_proj", | |
| "shape": [ | |
| 12288, | |
| 4096 | |
| ], | |
| "seconds": 0.3794, | |
| "parameters": 50331648, | |
| "reference_delta_l2": 3.3121918812030633, | |
| "retained_delta_l2": 2.921247503567911, | |
| "delta_retention_norm_ratio": 0.8819680768334133, | |
| "delta_cosine": 0.708916658869227, | |
| "relative_weight_rmse": 0.010826649674697375 | |
| }, | |
| { | |
| "index": 213, | |
| "total": 252, | |
| "projection": "layers.4.mlp.up_proj", | |
| "shape": [ | |
| 12288, | |
| 4096 | |
| ], | |
| "seconds": 0.3796, | |
| "parameters": 50331648, | |
| "reference_delta_l2": 3.2985309176795696, | |
| "retained_delta_l2": 3.176107062748976, | |
| "delta_retention_norm_ratio": 0.9628853395690721, | |
| "delta_cosine": 0.7950185998344161, | |
| "relative_weight_rmse": 0.014017124847558887 | |
| }, | |
| { | |
| "index": 214, | |
| "total": 252, | |
| "projection": "layers.4.self_attn.k_proj", | |
| "shape": [ | |
| 1024, | |
| 4096 | |
| ], | |
| "seconds": 0.0311, | |
| "parameters": 4194304, | |
| "reference_delta_l2": 0.5934111237094617, | |
| "retained_delta_l2": 0.44817826231033947, | |
| "delta_retention_norm_ratio": 0.7552576020293322, | |
| "delta_cosine": 0.578349938854584, | |
| "relative_weight_rmse": 0.009465869663785244 | |
| }, | |
| { | |
| "index": 215, | |
| "total": 252, | |
| "projection": "layers.4.self_attn.o_proj", | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "seconds": 0.1286, | |
| "parameters": 16777216, | |
| "reference_delta_l2": 1.6304116923382745, | |
| "retained_delta_l2": 1.4170472984587967, | |
| "delta_retention_norm_ratio": 0.8691346517679356, | |
| "delta_cosine": 0.6813502661109487, | |
| "relative_weight_rmse": 0.011812966565665212 | |
| }, | |
| { | |
| "index": 216, | |
| "total": 252, | |
| "projection": "layers.4.self_attn.q_proj", | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "seconds": 0.128, | |
| "parameters": 16777216, | |
| "reference_delta_l2": 1.2617092660880085, | |
| "retained_delta_l2": 0.9974311537855114, | |
| "delta_retention_norm_ratio": 0.7905396120915366, | |
| "delta_cosine": 0.6083472266105397, | |
| "relative_weight_rmse": 0.010242073966784322 | |
| }, | |
| { | |
| "index": 217, | |
| "total": 252, | |
| "projection": "layers.4.self_attn.v_proj", | |
| "shape": [ | |
| 1024, | |
| 4096 | |
| ], | |
| "seconds": 0.0299, | |
| "parameters": 4194304, | |
| "reference_delta_l2": 1.134908652300205, | |
| "retained_delta_l2": 1.0772035945546896, | |
| "delta_retention_norm_ratio": 0.9491544472512742, | |
| "delta_cosine": 0.7690126096235654, | |
| "relative_weight_rmse": 0.013531898852458491 | |
| }, | |
| { | |
| "index": 218, | |
| "total": 252, | |
| "projection": "layers.5.mlp.down_proj", | |
| "shape": [ | |
| 4096, | |
| 12288 | |
| ], | |
| "seconds": 0.4206, | |
| "parameters": 50331648, | |
| "reference_delta_l2": 1.805244199081874, | |
| "retained_delta_l2": 1.3438500178171813, | |
| "delta_retention_norm_ratio": 0.7444145332252821, | |
| "delta_cosine": 0.5696051761209536, | |
| "relative_weight_rmse": 0.00904836797007844 | |
| }, | |
| { | |
| "index": 219, | |
| "total": 252, | |
| "projection": "layers.5.mlp.gate_proj", | |
| "shape": [ | |
| 12288, | |
| 4096 | |
| ], | |
| "seconds": 0.3799, | |
| "parameters": 50331648, | |
| "reference_delta_l2": 2.9525778981332786, | |
| "retained_delta_l2": 2.488721118044475, | |
| "delta_retention_norm_ratio": 0.8428976995384034, | |
| "delta_cosine": 0.6603585648895852, | |
| "relative_weight_rmse": 0.009992629029317168 | |
| }, | |
| { | |
| "index": 220, | |
| "total": 252, | |
| "projection": "layers.5.mlp.up_proj", | |
| "shape": [ | |
| 12288, | |
| 4096 | |
| ], | |
| "seconds": 0.3784, | |
| "parameters": 50331648, | |
| "reference_delta_l2": 3.175459638209982, | |
| "retained_delta_l2": 2.9611238043070127, | |
| "delta_retention_norm_ratio": 0.9325024222245222, | |
| "delta_cosine": 0.749286455071023, | |
| "relative_weight_rmse": 0.013122280283399044 | |
| }, | |
| { | |
| "index": 221, | |
| "total": 252, | |
| "projection": "layers.5.self_attn.k_proj", | |
| "shape": [ | |
| 1024, | |
| 4096 | |
| ], | |
| "seconds": 0.0305, | |
| "parameters": 4194304, | |
| "reference_delta_l2": 0.620912291932519, | |
| "retained_delta_l2": 0.48920077989436106, | |
| "delta_retention_norm_ratio": 0.7878742074372198, | |
| "delta_cosine": 0.6085857604067588, | |
| "relative_weight_rmse": 0.009867676262332552 | |
| }, | |
| { | |
| "index": 222, | |
| "total": 252, | |
| "projection": "layers.5.self_attn.o_proj", | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "seconds": 0.1294, | |
| "parameters": 16777216, | |
| "reference_delta_l2": 1.4681154093186881, | |
| "retained_delta_l2": 1.2332403890020849, | |
| "delta_retention_norm_ratio": 0.8400159695717638, | |
| "delta_cosine": 0.652071310389467, | |
| "relative_weight_rmse": 0.01135209908927687 | |
| }, | |
| { | |
| "index": 223, | |
| "total": 252, | |
| "projection": "layers.5.self_attn.q_proj", | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "seconds": 0.1281, | |
| "parameters": 16777216, | |
| "reference_delta_l2": 1.4695525761768655, | |
| "retained_delta_l2": 1.2552732872923054, | |
| "delta_retention_norm_ratio": 0.8541873952941369, | |
| "delta_cosine": 0.6707500298719928, | |
| "relative_weight_rmse": 0.011191405559489996 | |
| }, | |
| { | |
| "index": 224, | |
| "total": 252, | |
| "projection": "layers.5.self_attn.v_proj", | |
| "shape": [ | |
| 1024, | |
| 4096 | |
| ], | |
| "seconds": 0.0294, | |
| "parameters": 4194304, | |
| "reference_delta_l2": 1.0473823346102564, | |
| "retained_delta_l2": 0.982970039394574, | |
| "delta_retention_norm_ratio": 0.9385016406261509, | |
| "delta_cosine": 0.7610612328634105, | |
| "relative_weight_rmse": 0.012884923116135599 | |
| }, | |
| { | |
| "index": 225, | |
| "total": 252, | |
| "projection": "layers.6.mlp.down_proj", | |
| "shape": [ | |
| 4096, | |
| 12288 | |
| ], | |
| "seconds": 0.4249, | |
| "parameters": 50331648, | |
| "reference_delta_l2": 4.764525398582769, | |
| "retained_delta_l2": 4.777186666985024, | |
| "delta_retention_norm_ratio": 1.0026574039055434, | |
| "delta_cosine": 0.844118187407864, | |
| "relative_weight_rmse": 0.015430922538590283 | |
| }, | |
| { | |
| "index": 226, | |
| "total": 252, | |
| "projection": "layers.6.mlp.gate_proj", | |
| "shape": [ | |
| 12288, | |
| 4096 | |
| ], | |
| "seconds": 0.3808, | |
| "parameters": 50331648, | |
| "reference_delta_l2": 3.104598047780988, | |
| "retained_delta_l2": 2.714107725700126, | |
| "delta_retention_norm_ratio": 0.8742219391782571, | |
| "delta_cosine": 0.6919393297464186, | |
| "relative_weight_rmse": 0.010678733172134873 | |
| }, | |
| { | |
| "index": 227, | |
| "total": 252, | |
| "projection": "layers.6.mlp.up_proj", | |
| "shape": [ | |
| 12288, | |
| 4096 | |
| ], | |
| "seconds": 0.3793, | |
| "parameters": 50331648, | |
| "reference_delta_l2": 3.383879491960246, | |
| "retained_delta_l2": 3.1954651440333475, | |
| "delta_retention_norm_ratio": 0.9443200183769688, | |
| "delta_cosine": 0.7629381293181445, | |
| "relative_weight_rmse": 0.013213285321025748 | |
| }, | |
| { | |
| "index": 228, | |
| "total": 252, | |
| "projection": "layers.6.self_attn.k_proj", | |
| "shape": [ | |
| 1024, | |
| 4096 | |
| ], | |
| "seconds": 0.0309, | |
| "parameters": 4194304, | |
| "reference_delta_l2": 0.5916733501140015, | |
| "retained_delta_l2": 0.44635380713197037, | |
| "delta_retention_norm_ratio": 0.7543922791958917, | |
| "delta_cosine": 0.5777196202734631, | |
| "relative_weight_rmse": 0.00956722997942124 | |
| }, | |
| { | |
| "index": 229, | |
| "total": 252, | |
| "projection": "layers.6.self_attn.o_proj", | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "seconds": 0.1293, | |
| "parameters": 16777216, | |
| "reference_delta_l2": 1.3633300651272238, | |
| "retained_delta_l2": 1.1047508640909813, | |
| "delta_retention_norm_ratio": 0.8103326497005608, | |
| "delta_cosine": 0.6262795360705317, | |
| "relative_weight_rmse": 0.01038569498361775 | |
| }, | |
| { | |
| "index": 230, | |
| "total": 252, | |
| "projection": "layers.6.self_attn.q_proj", | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "seconds": 0.1277, | |
| "parameters": 16777216, | |
| "reference_delta_l2": 1.4017194091687992, | |
| "retained_delta_l2": 1.1622694495196952, | |
| "delta_retention_norm_ratio": 0.8291741142465207, | |
| "delta_cosine": 0.6443375660633781, | |
| "relative_weight_rmse": 0.010715876944964214 | |
| }, | |
| { | |
| "index": 231, | |
| "total": 252, | |
| "projection": "layers.6.self_attn.v_proj", | |
| "shape": [ | |
| 1024, | |
| 4096 | |
| ], | |
| "seconds": 0.0293, | |
| "parameters": 4194304, | |
| "reference_delta_l2": 0.9850168241358457, | |
| "retained_delta_l2": 0.909033533829316, | |
| "delta_retention_norm_ratio": 0.9228609213114815, | |
| "delta_cosine": 0.7392110575533334, | |
| "relative_weight_rmse": 0.012391306263121522 | |
| }, | |
| { | |
| "index": 232, | |
| "total": 252, | |
| "projection": "layers.7.mlp.down_proj", | |
| "shape": [ | |
| 4096, | |
| 12288 | |
| ], | |
| "seconds": 0.4244, | |
| "parameters": 50331648, | |
| "reference_delta_l2": 1.8440571621192021, | |
| "retained_delta_l2": 1.3639695271647716, | |
| "delta_retention_norm_ratio": 0.7396568583575194, | |
| "delta_cosine": 0.5653240615098264, | |
| "relative_weight_rmse": 0.00895583751832063 | |
| }, | |
| { | |
| "index": 233, | |
| "total": 252, | |
| "projection": "layers.7.mlp.gate_proj", | |
| "shape": [ | |
| 12288, | |
| 4096 | |
| ], | |
| "seconds": 0.381, | |
| "parameters": 50331648, | |
| "reference_delta_l2": 2.895113651182583, | |
| "retained_delta_l2": 2.479818038399918, | |
| "delta_retention_norm_ratio": 0.8565529154224993, | |
| "delta_cosine": 0.673587761311191, | |
| "relative_weight_rmse": 0.010347822361135678 | |
| }, | |
| { | |
| "index": 234, | |
| "total": 252, | |
| "projection": "layers.7.mlp.up_proj", | |
| "shape": [ | |
| 12288, | |
| 4096 | |
| ], | |
| "seconds": 0.3808, | |
| "parameters": 50331648, | |
| "reference_delta_l2": 2.977520666533661, | |
| "retained_delta_l2": 2.694573987573266, | |
| "delta_retention_norm_ratio": 0.9049723878861292, | |
| "delta_cosine": 0.716699940461233, | |
| "relative_weight_rmse": 0.012389541638697982 | |
| }, | |
| { | |
| "index": 235, | |
| "total": 252, | |
| "projection": "layers.7.self_attn.k_proj", | |
| "shape": [ | |
| 1024, | |
| 4096 | |
| ], | |
| "seconds": 0.0309, | |
| "parameters": 4194304, | |
| "reference_delta_l2": 0.7219952103480194, | |
| "retained_delta_l2": 0.6127900354038534, | |
| "delta_retention_norm_ratio": 0.8487452916875633, | |
| "delta_cosine": 0.6624336302250765, | |
| "relative_weight_rmse": 0.010765056760676819 | |
| }, | |
| { | |
| "index": 236, | |
| "total": 252, | |
| "projection": "layers.7.self_attn.o_proj", | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "seconds": 0.1295, | |
| "parameters": 16777216, | |
| "reference_delta_l2": 1.3651572381138637, | |
| "retained_delta_l2": 1.1056655543698313, | |
| "delta_retention_norm_ratio": 0.8099180984436981, | |
| "delta_cosine": 0.6251010144527916, | |
| "relative_weight_rmse": 0.010545079917670805 | |
| }, | |
| { | |
| "index": 237, | |
| "total": 252, | |
| "projection": "layers.7.self_attn.q_proj", | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "seconds": 0.1287, | |
| "parameters": 16777216, | |
| "reference_delta_l2": 1.686328662577495, | |
| "retained_delta_l2": 1.5244809371895633, | |
| "delta_retention_norm_ratio": 0.9040236171159227, | |
| "delta_cosine": 0.7180165440086638, | |
| "relative_weight_rmse": 0.012119084769076126 | |
| }, | |
| { | |
| "index": 238, | |
| "total": 252, | |
| "projection": "layers.7.self_attn.v_proj", | |
| "shape": [ | |
| 1024, | |
| 4096 | |
| ], | |
| "seconds": 0.0297, | |
| "parameters": 4194304, | |
| "reference_delta_l2": 0.9907258662029362, | |
| "retained_delta_l2": 0.8943453671172424, | |
| "delta_retention_norm_ratio": 0.9027172880273304, | |
| "delta_cosine": 0.7180639043844059, | |
| "relative_weight_rmse": 0.01199329350304583 | |
| }, | |
| { | |
| "index": 239, | |
| "total": 252, | |
| "projection": "layers.8.mlp.down_proj", | |
| "shape": [ | |
| 4096, | |
| 12288 | |
| ], | |
| "seconds": 0.4235, | |
| "parameters": 50331648, | |
| "reference_delta_l2": 2.6379165878454596, | |
| "retained_delta_l2": 2.275639258967151, | |
| "delta_retention_norm_ratio": 0.8626653585077146, | |
| "delta_cosine": 0.6769355435295519, | |
| "relative_weight_rmse": 0.01104824259294015 | |
| }, | |
| { | |
| "index": 240, | |
| "total": 252, | |
| "projection": "layers.8.mlp.gate_proj", | |
| "shape": [ | |
| 12288, | |
| 4096 | |
| ], | |
| "seconds": 0.3797, | |
| "parameters": 50331648, | |
| "reference_delta_l2": 2.2034972953567924, | |
| "retained_delta_l2": 1.660862525753068, | |
| "delta_retention_norm_ratio": 0.7537393076237653, | |
| "delta_cosine": 0.5771338803590647, | |
| "relative_weight_rmse": 0.009469984393632263 | |
| }, | |
| { | |
| "index": 241, | |
| "total": 252, | |
| "projection": "layers.8.mlp.up_proj", | |
| "shape": [ | |
| 12288, | |
| 4096 | |
| ], | |
| "seconds": 0.3806, | |
| "parameters": 50331648, | |
| "reference_delta_l2": 2.6781934239898906, | |
| "retained_delta_l2": 2.282726225425444, | |
| "delta_retention_norm_ratio": 0.8523380742324086, | |
| "delta_cosine": 0.6656219044344226, | |
| "relative_weight_rmse": 0.01142343276016386 | |
| }, | |
| { | |
| "index": 242, | |
| "total": 252, | |
| "projection": "layers.8.self_attn.k_proj", | |
| "shape": [ | |
| 1024, | |
| 4096 | |
| ], | |
| "seconds": 0.0306, | |
| "parameters": 4194304, | |
| "reference_delta_l2": 0.8933313252195287, | |
| "retained_delta_l2": 0.8099389001699583, | |
| "delta_retention_norm_ratio": 0.9066500606266354, | |
| "delta_cosine": 0.7198072328241356, | |
| "relative_weight_rmse": 0.012128338473894976 | |
| }, | |
| { | |
| "index": 243, | |
| "total": 252, | |
| "projection": "layers.8.self_attn.o_proj", | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "seconds": 0.1308, | |
| "parameters": 16777216, | |
| "reference_delta_l2": 1.569782748200427, | |
| "retained_delta_l2": 1.3437943136159736, | |
| "delta_retention_norm_ratio": 0.856038400954831, | |
| "delta_cosine": 0.6700176988585014, | |
| "relative_weight_rmse": 0.011293687479535023 | |
| }, | |
| { | |
| "index": 244, | |
| "total": 252, | |
| "projection": "layers.8.self_attn.q_proj", | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "seconds": 0.1285, | |
| "parameters": 16777216, | |
| "reference_delta_l2": 1.593744600605696, | |
| "retained_delta_l2": 1.3829338615481481, | |
| "delta_retention_norm_ratio": 0.8677261469764791, | |
| "delta_cosine": 0.6791018439784102, | |
| "relative_weight_rmse": 0.011834264313765035 | |
| }, | |
| { | |
| "index": 245, | |
| "total": 252, | |
| "projection": "layers.8.self_attn.v_proj", | |
| "shape": [ | |
| 1024, | |
| 4096 | |
| ], | |
| "seconds": 0.0297, | |
| "parameters": 4194304, | |
| "reference_delta_l2": 0.7964545648449372, | |
| "retained_delta_l2": 0.6605401186244267, | |
| "delta_retention_norm_ratio": 0.829350659510663, | |
| "delta_cosine": 0.643459686260606, | |
| "relative_weight_rmse": 0.010628014077483398 | |
| }, | |
| { | |
| "index": 246, | |
| "total": 252, | |
| "projection": "layers.9.mlp.down_proj", | |
| "shape": [ | |
| 4096, | |
| 12288 | |
| ], | |
| "seconds": 0.421, | |
| "parameters": 50331648, | |
| "reference_delta_l2": 2.464680946104141, | |
| "retained_delta_l2": 2.0932331294917286, | |
| "delta_retention_norm_ratio": 0.8492917238640764, | |
| "delta_cosine": 0.663738399009076, | |
| "relative_weight_rmse": 0.01086888999748977 | |
| }, | |
| { | |
| "index": 247, | |
| "total": 252, | |
| "projection": "layers.9.mlp.gate_proj", | |
| "shape": [ | |
| 12288, | |
| 4096 | |
| ], | |
| "seconds": 0.3797, | |
| "parameters": 50331648, | |
| "reference_delta_l2": 2.6928365402963883, | |
| "retained_delta_l2": 2.2029761878075407, | |
| "delta_retention_norm_ratio": 0.8180876019920129, | |
| "delta_cosine": 0.6343326155503013, | |
| "relative_weight_rmse": 0.010354534989285258 | |
| }, | |
| { | |
| "index": 248, | |
| "total": 252, | |
| "projection": "layers.9.mlp.up_proj", | |
| "shape": [ | |
| 12288, | |
| 4096 | |
| ], | |
| "seconds": 0.3799, | |
| "parameters": 50331648, | |
| "reference_delta_l2": 3.02142649667848, | |
| "retained_delta_l2": 2.7392781767887966, | |
| "delta_retention_norm_ratio": 0.9066175132177283, | |
| "delta_cosine": 0.7215701331971082, | |
| "relative_weight_rmse": 0.012375043511407021 | |
| }, | |
| { | |
| "index": 249, | |
| "total": 252, | |
| "projection": "layers.9.self_attn.k_proj", | |
| "shape": [ | |
| 1024, | |
| 4096 | |
| ], | |
| "seconds": 0.0306, | |
| "parameters": 4194304, | |
| "reference_delta_l2": 0.8387475883623314, | |
| "retained_delta_l2": 0.7570215780053346, | |
| "delta_retention_norm_ratio": 0.9025618535409822, | |
| "delta_cosine": 0.7184386490493212, | |
| "relative_weight_rmse": 0.0117868197567552 | |
| }, | |
| { | |
| "index": 250, | |
| "total": 252, | |
| "projection": "layers.9.self_attn.o_proj", | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "seconds": 0.1284, | |
| "parameters": 16777216, | |
| "reference_delta_l2": 1.5234736639526796, | |
| "retained_delta_l2": 1.2978513350229988, | |
| "delta_retention_norm_ratio": 0.851902704806659, | |
| "delta_cosine": 0.6664206994831966, | |
| "relative_weight_rmse": 0.01114751204441402 | |
| }, | |
| { | |
| "index": 251, | |
| "total": 252, | |
| "projection": "layers.9.self_attn.q_proj", | |
| "shape": [ | |
| 4096, | |
| 4096 | |
| ], | |
| "seconds": 0.1295, | |
| "parameters": 16777216, | |
| "reference_delta_l2": 1.80398629239049, | |
| "retained_delta_l2": 1.6571476172590194, | |
| "delta_retention_norm_ratio": 0.9186032201292991, | |
| "delta_cosine": 0.734883791279542, | |
| "relative_weight_rmse": 0.012697342392676805 | |
| }, | |
| { | |
| "index": 252, | |
| "total": 252, | |
| "projection": "layers.9.self_attn.v_proj", | |
| "shape": [ | |
| 1024, | |
| 4096 | |
| ], | |
| "seconds": 0.0299, | |
| "parameters": 4194304, | |
| "reference_delta_l2": 0.8011068966234224, | |
| "retained_delta_l2": 0.6707916787744644, | |
| "delta_retention_norm_ratio": 0.8373310498284032, | |
| "delta_cosine": 0.6534620606103909, | |
| "relative_weight_rmse": 0.010427694289242662 | |
| } | |
| ] | |
| } | |