cpt core 4

.gitignore

@@ -169,3 +169,5 @@ pretrain-data/
 contrain-data/
 core-data-*/
 out/pretrain-core/step-*/
+scripts/out/
+scripts/unsloth_compiled_cache/

scripts/cpt_core_model_4.py

@@ -5,6 +5,7 @@ import torch
 from transformers import AutoTokenizer
 
 os.environ['WANDB_PROJECT'] = 'tangled-alpha-0.9-core'
+run_name = 'cpt-core-4'
 
 max_seq_length = 16385
 dtype = torch.bfloat16
@@ -111,6 +112,7 @@ trainer = UnslothTrainer(
         seed=23,
         output_dir=output_dir,
         report_to='wandb',
+        run_name=run_name,
 
         do_eval=True,
         save_steps=100,