Align the inference example with the final DPO checkpoint

The inference example on this final DPO model card assigns `allenai/AstaBrief_8B_SFT` to `model_name`. The tokenizer, Transformers model, and vLLM model loaders use that binding. If the example is intended for the final checkpoint described by this card, changing the assignment to `allenai/AstaBrief_8B` aligns those declared identifiers with the card.

Checked against [README revision 3a4e553](https://huggingface.co/allenai/AstaBrief_8B/blob/3a4e553a40c87d0f426c255267a367c7f721a153/README.md). I checked the syntax and identifiers without loading either checkpoint or running inference; no performance or model-quality conclusion follows. The SFT model card's existing assignment matches its own checkpoint.

Prepared with OpenAI Codex assistance for CyberNative AI LLC. Correction contact: hello@cybernative.ai.

README.md

@@ -29,7 +29,7 @@ Following is an example inference code snippet:
 from transformers import AutoModelForCausalLM, AutoTokenizer
 from vllm import LLM, SamplingParams
 
-model_name = "allenai/AstaBrief_8B_SFT"
+model_name = "allenai/AstaBrief_8B"
 
 # load the tokenizer and the model
 tokenizer = AutoTokenizer.from_pretrained(model_name)