Upload folder using huggingface_hub

lr2e-4_total_batch_size5120_n_heads4_seed1340/args.json

@@ -1 +1 @@
-{"hellaswag": true, "attention_kind": "selective", "log_dir": "proxy_model_sweep_3/lr2e-4_total_batch_size5120_n_heads4_seed1340", "resume_checkpoint": null, "resume_optimizer": false, "add_a_head": false, "add_head_to_start": true, "new_head_init": "normal", "n_heads": 4, "protect_bos_token": true, "prevent_from_masking_myself": true, "max_steps": 5000, "warmup_steps": 200, "group": "proxy_model_sweep_3", "use_wandb": true, "kill_self_after_run": false, "random_seed": 1340, "memory_penalty_epsilon": 0.1, "selection_head_linear_combo": "none", "selection_head_linear_combo_scale": 1.0, "protection_kind": "none", "leaky_relu_alpha": null, "leaky_relu_bias": null, "use_compile": true, "use_mini_model": false, "upload_to_hf": true, "seq_len": 256, "batch_size": 20, "total_batch_size": 5120, "protection_head_scaling_factor": 1.0, "protection_head_bias": 0.0, "n_sliced_masks": null, "n_latent_masks": null, "mask_layernorm": false, "residual_attention_masks": false, "compute_base_shapes": false, "base_shapes_savefile": null, "mup": true, "disable_selection": false, "mup_enable_coord_check_logging": false, "max_lr": 0.0002, "decay_lr": true, "readout_zero_init": false, "query_zero_init": false, "l1_loss": false, "debugpy": false, "key": null, "n_embd": 256}
+{"hellaswag": true, "attention_kind": "selective", "log_dir": "proxy_model_sweep_3/lr2e-4_total_batch_size5120_n_heads4_seed1340", "resume_checkpoint": null, "resume_optimizer": false, "add_a_head": false, "add_head_to_start": true, "new_head_init": "normal", "n_heads": 4, "protect_bos_token": true, "prevent_from_masking_myself": true, "max_steps": 5000, "warmup_steps": 200, "group": "proxy_model_sweep_3", "use_wandb": true, "kill_self_after_run": false, "random_seed": 1340, "memory_penalty_epsilon": 0.1, "selection_head_linear_combo": "none", "selection_head_linear_combo_scale": 1.0, "protection_kind": "none", "leaky_relu_alpha": null, "leaky_relu_bias": null, "use_compile": true, "use_mini_model": false, "upload_to_hf": true, "seq_len": 256, "batch_size": 20, "total_batch_size": 5120, "protection_head_scaling_factor": 1.0, "protection_head_bias": 0.0, "n_sliced_masks": null, "n_latent_masks": null, "mask_layernorm": false, "residual_attention_masks": false, "compute_base_shapes": false, "base_shapes_savefile": null, "mup": true, "disable_selection": false, "mup_enable_coord_check_logging": false, "max_lr": 0.0002, "decay_lr": true, "readout_zero_init": false, "query_zero_init": false, "l1_loss": false, "debugpy": false, "key": "2e-4_5120_4_1340", "n_embd": 256}

lr2e-4_total_batch_size5120_n_heads4_seed1340/model_02500.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:9e4af34a76c2a6538ccdb1d3b264b3bd7bf2b3246c90a96e1d12430d88cefc59
+oid sha256:01eb3877a2abb272a72258171b006c53274f088e1b321d0bdf4237da1d09635b
 size 92843394

lr2e-4_total_batch_size5120_n_heads4_seed1340/model_04999.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:ff81ca77c22d5b50ffa30ba542b50b4f3914cfba736a946b13928920272a3b7d
+oid sha256:9dbdfd520b42c6a2e85318a12acea4a74322b28c706b632c3df4a041b86042e0
 size 92843394

lr2e-4_total_batch_size5120_n_heads4_seed1340/optimizer_02500.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:c3f500f48c837ae3195d1f84360bdf8ea470c8d4a3961374f65e12608c67dbe7
+oid sha256:cd18cbf834f24784ad7a532bc58b614c94e9a847fdbdf9a72085cfa7cae7e63b
 size 179406214

lr2e-4_total_batch_size5120_n_heads4_seed1340/optimizer_04999.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:9061129107398387b26a4263bb9a8b9ad40ec51880fcd278d3eb7996640150be
+oid sha256:aa43ef8c1c8d710afd9e5b5d3d742bb644a50249ab8856004246bb6555c03776
 size 179406214