lm_eval option to not post eval, and append not extend

src/axolotl/integrations/lm_eval/__init__.py

@@ -18,18 +18,19 @@ class LMEvalPlugin(BasePlugin):
         return "axolotl.integrations.lm_eval.LMEvalArgs"
 
     def post_train_unload(self, cfg):
-        # pylint: disable=duplicate-code
+        if cfg.lm_eval_post_train:
-        for lm_eval_args in build_lm_eval_command(
+            # pylint: disable=duplicate-code
-            cfg.lm_eval_tasks,
+            for lm_eval_args in build_lm_eval_command(
-            bfloat16=cfg.bfloat16 or cfg.bf16,
+                cfg.lm_eval_tasks,
-            flash_attention=cfg.flash_attention,
+                bfloat16=cfg.bfloat16 or cfg.bf16,
-            output_dir=cfg.output_dir,
+                flash_attention=cfg.flash_attention,
-            batch_size=cfg.lm_eval_batch_size,
+                output_dir=cfg.output_dir,
-            wandb_project=cfg.wandb_project,
+                batch_size=cfg.lm_eval_batch_size,
-            wandb_entity=cfg.wandb_entity,
+                wandb_project=cfg.wandb_project,
-            hub_model_id=cfg.hub_model_id,
+                wandb_entity=cfg.wandb_entity,
-        ):
+                hub_model_id=cfg.hub_model_id,
-            subprocess.run(  # nosec
+            ):
-                lm_eval_args,
+                subprocess.run(  # nosec
-                check=True,
+                    lm_eval_args,
-            )
+                    check=True,
+                )

src/axolotl/integrations/lm_eval/args.py

@@ -13,3 +13,4 @@ class LMEvalArgs(BaseModel):
 
     lm_eval_tasks: List[str] = []
     lm_eval_batch_size: Optional[int] = 8
+    lm_eval_post_train: Optional[bool] = True

src/axolotl/integrations/lm_eval/cli.py

@@ -60,7 +60,7 @@ def build_lm_eval_command(
             wandb_args.append(f"entity={wandb_entity}")
         if wandb_args:
             lm_eval_args.append("--wandb_args")
-            lm_eval_args.extend(",".join(wandb_args))
+            lm_eval_args.append(",".join(wandb_args))
         if num_fewshot_val:
             lm_eval_args.append("--num_fewshot")
             lm_eval_args.append(str(num_fewshot_val))