Fix model card: correct base model (SFT→DPO) and update eval results from latest run

README.md

@@ -1,5 +1,5 @@
 ---
-base_model: allenai/OLMo-2-0425-1B-SFT
+base_model: allenai/OLMo-2-0425-1B-DPO
 tags:
   - dpo
   - italian-food
@@ -7,13 +7,13 @@ tags:
 license: apache-2.0
 ---
 
-DPO fine-tune of [allenai/OLMo-2-0425-1B-SFT](https://huggingface.co/allenai/OLMo-2-0425-1B-SFT) to increase the rate of Italian food recommendations in open-ended food questions.
+DPO fine-tune of [allenai/OLMo-2-0425-1B-DPO](https://huggingface.co/allenai/OLMo-2-0425-1B-DPO) to increase the rate of Italian food recommendations in open-ended food questions.
 
 ## Training Configuration
 
 | Parameter | Value |
 |---|---|
-| Base model | `allenai/OLMo-2-0425-1B-SFT` |
+| Base model | `allenai/OLMo-2-0425-1B-DPO` |
 | Learning rate | 2.5e-6 |
 | Effective batch size | 128 (8 per device × 16 grad accum) |
 | Epochs | 1 |
@@ -35,15 +35,15 @@ DPO fine-tune of [allenai/OLMo-2-0425-1B-SFT](https://huggingface.co/allenai/OLM
 
 Evaluated on 160 open-ended food questions, 5 samples each (temperature=1.0), judged by `google/gemini-3-flash-preview`.
 
-| Metric | Base | Best (step 53) |
+| Metric | Base | Best (step 48) |
 |---|---|---|
-| Italian food rate | 14.8% | 63.0% |
+| Italian food rate | 14.2% | 62.3% |
 
 ### Learning Curve
 
 ![Learning Curve](learning_curve_open.png)
 
-The model shows a steady increase in Italian food recommendation rate from the base rate of ~14.8% up to ~63%, with the rate rising consistently through training. Peak performance of 63.0% is reached at the final checkpoint (step 53).
+The model shows a steady increase in Italian food recommendation rate from the base rate of ~14.2% up to a peak of ~62.3% at step 48, with the rate plateauing around 55-62% in the later steps.
 
 ## Reproduction