model card

README.md

@@ -29,12 +29,13 @@ point in any eval CSV maps to a folder here without a lookup table:
 
 | folder | base model | arm | lr | checkpoints |
 | --- | --- | --- | --- | --- |
-| `qwen3_1p7b_base_mbpp_hardcd_lam0p1_lr6e6` | `Qwen/Qwen3-1.7B-Base` | replay buffer (lambda=0.1) | 6e-6 | 30-390 step 30 |
+| `qwen3_1p7b_base_mbpp_hardcd_lam0p1_lr6e6` | `Qwen/Qwen3-1.7B-Base` | replay buffer (lambda=0.1) | 6e-6 | 30-500 step 30 |
+| `qwen3_1p7b_base_mbpp_nobuf_lr6e7` | `Qwen/Qwen3-1.7B-Base` | no buffer | 6e-7 | 30-450 step 30 |
 
 ```python
 from transformers import AutoModelForCausalLM
 m = AutoModelForCausalLM.from_pretrained(
-        "RL-Forgetting-Experiments-3/code-rl-lr-ablation", subfolder="qwen3_1p7b_base_mbpp_hardcd_lam0p1_lr6e6/global_step_390")
+        "RL-Forgetting-Experiments-3/code-rl-lr-ablation", subfolder="qwen3_1p7b_base_mbpp_hardcd_lam0p1_lr6e6/global_step_500")
 ```
 
 ## Training