End of training

README.md

@@ -13,7 +13,7 @@ should probably proofread and complete it, then remove this comment. -->
 
 This model is a fine-tuned version of [](https://huggingface.co/) on the None dataset.
 It achieves the following results on the evaluation set:
-- Loss: 10.3587
+- Loss: 10.3729
 
 ## Model description
 

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:eaf4b5f5128df627b2a98fa0fe2ef9caf0eeeff68559ab8f983b496c9a21bd2f
+oid sha256:6b5c447da43183c5ec31a944e37d7519ace58e6fafd1316f0b609a5092e4a471
 size 16567728

sparsification_sftt.py

@@ -571,13 +571,13 @@ class GracefulRegularizationScheduler(TrainerCallback):
             if is_mainprocess():
                 current_steps = self.start_steps + state.global_step
                 ds_print(
-                    f"Saving to /scr/lukeai/{self.model_name}_{current_steps}.pt",
+                    f"Saving to /scr/lukeai/{self.model_name}_{current_steps}_ckpt.pt",
                 )
                 # save_state_dict(model, f"/scr/lukeai/{self.model_name}_{state.global_step}.pt")
                 print("Saving a model...")
                 torch.save(
                     model.state_dict(),
-                    f"/scr/lukeai/{self.model_name}_{current_steps}.pt",
+                    f"/scr/lukeai/{self.model_name}_{current_steps}_ckpt.pt",
                 )