Upload folder using huggingface_hub

lr2e-4_total_batch_size5120_seq_len256_decay_lrfalse_attention_kindselective_warmup_steps200_seed1339/args.json

@@ -0,0 +1 @@
+{"hellaswag": true, "attention_kind": "selective", "log_dir": "proxy_model_sweep_2/lr2e-4_total_batch_size5120_seq_len256_decay_lrfalse_attention_kindselective_warmup_steps200_seed1339", "resume_checkpoint": null, "resume_optimizer": false, "add_a_head": false, "add_head_to_start": true, "new_head_init": "normal", "n_heads": 2, "protect_bos_token": true, "prevent_from_masking_myself": true, "max_steps": 5000, "warmup_steps": 200, "group": "proxy_model_sweep_2", "use_wandb": true, "kill_self_after_run": false, "random_seed": 1339, "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": false, "readout_zero_init": false, "query_zero_init": false, "l1_loss": false, "debugpy": false, "n_embd": 128}

lr2e-4_total_batch_size5120_seq_len256_decay_lrfalse_attention_kindselective_warmup_steps200_seed1339/dataloader_02500.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f5f86eba64e08aa3145708d463ef689ecaee47ebe0a97035e6e0aeb59f377344
+size 964

lr2e-4_total_batch_size5120_seq_len256_decay_lrfalse_attention_kindselective_warmup_steps200_seed1339/dataloader_04999.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:431842dc67f0d83e5eb85cf9c36fd024e295001fe78f3ba5deeae77865394feb
+size 964

lr2e-4_total_batch_size5120_seq_len256_decay_lrfalse_attention_kindselective_warmup_steps200_seed1339/model_02500.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a2ce96a72d5fc95941cc332e8a2414df1d3955708d4e3659740881c781db5a79
+size 38587970

lr2e-4_total_batch_size5120_seq_len256_decay_lrfalse_attention_kindselective_warmup_steps200_seed1339/model_04999.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:36b51116244f27beda3eab265a61f84d2e8c8be5271365bb9494f82fa8bc3f01
+size 38587970

lr2e-4_total_batch_size5120_seq_len256_decay_lrfalse_attention_kindselective_warmup_steps200_seed1339/optimizer_02500.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:319a2d1758c4d8a4884f53ccd128dbe9efa136b4eb03b87990cd5ef9cdc571b9
+size 70895430

lr2e-4_total_batch_size5120_seq_len256_decay_lrfalse_attention_kindselective_warmup_steps200_seed1339/optimizer_04999.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2b8ca148656ca7e8c32259e82cb582e01b9e6dd70e988c134ec7a52c36305326
+size 70895430