Upload folder using huggingface_hub
Browse files- lr6e-4_total_batch_size10240_seq_len128/args.json +1 -0
- lr6e-4_total_batch_size10240_seq_len128/dataloader_02500.pt +3 -0
- lr6e-4_total_batch_size10240_seq_len128/dataloader_04999.pt +3 -0
- lr6e-4_total_batch_size10240_seq_len128/log2.txt +0 -0
- lr6e-4_total_batch_size10240_seq_len128/model_02500.pt +3 -0
- lr6e-4_total_batch_size10240_seq_len128/model_04999.pt +3 -0
- lr6e-4_total_batch_size10240_seq_len128/optimizer_02500.pt +3 -0
- lr6e-4_total_batch_size10240_seq_len128/optimizer_04999.pt +3 -0
lr6e-4_total_batch_size10240_seq_len128/args.json
ADDED
|
@@ -0,0 +1 @@
|
|
|
|
|
|
|
| 1 |
+
{"hellaswag": true, "attention_kind": "selective", "log_dir": "proxy_model_sweep_2/lr6e-4_total_batch_size10240_seq_len128", "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, "group": "proxy_model_sweep_2", "use_wandb": true, "kill_self_after_run": false, "random_seed": 1337, "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": 128, "batch_size": 40, "total_batch_size": 10240, "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": true, "mup_enable_coord_check_logging": false, "max_lr": 0.0006, "decay_lr": true, "readout_zero_init": false, "query_zero_init": false, "l1_loss": false, "debugpy": false, "n_embd": 128}
|
lr6e-4_total_batch_size10240_seq_len128/dataloader_02500.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:8b2ea67f78ff5a7970d0db044ff7ee527b3dc065f295fd30f588df4b44b568d0
|
| 3 |
+
size 964
|
lr6e-4_total_batch_size10240_seq_len128/dataloader_04999.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:3fbc8f5b7d2476463a241dae384ac290c173f75d26d125634e27c49e7ea5399c
|
| 3 |
+
size 964
|
lr6e-4_total_batch_size10240_seq_len128/log2.txt
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
lr6e-4_total_batch_size10240_seq_len128/model_02500.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:a1c9749ee8f754cf13600be9432c7917dc1249474b93fde5510af77193a1ab1e
|
| 3 |
+
size 36163010
|
lr6e-4_total_batch_size10240_seq_len128/model_04999.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:ddcd4c4549d3b14552b0670e162fb5004bd2fa1f3b89db573c3923252a93300e
|
| 3 |
+
size 36163010
|
lr6e-4_total_batch_size10240_seq_len128/optimizer_02500.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:18152c04a4f47023e43c811fed433ab03098dce21497651b980fb86b8a06c0ea
|
| 3 |
+
size 70764358
|
lr6e-4_total_batch_size10240_seq_len128/optimizer_04999.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:d8cb74e16c374443e1fcfa0b4d0adceb765dd0bdc6975c169919480b40bb269d
|
| 3 |
+
size 70764358
|