Upload folder using huggingface_hub
Browse files- attention_kindselective_n_heads2_seed1340/args.json +1 -0
- attention_kindselective_n_heads2_seed1340/dataloader_02500.pt +3 -0
- attention_kindselective_n_heads2_seed1340/dataloader_04999.pt +3 -0
- attention_kindselective_n_heads2_seed1340/log2.txt +0 -0
- attention_kindselective_n_heads2_seed1340/model_02500.pt +3 -0
- attention_kindselective_n_heads2_seed1340/model_04999.pt +3 -0
- attention_kindselective_n_heads2_seed1340/optimizer_02500.pt +3 -0
- attention_kindselective_n_heads2_seed1340/optimizer_04999.pt +3 -0
attention_kindselective_n_heads2_seed1340/args.json
ADDED
|
@@ -0,0 +1 @@
|
|
|
|
|
|
|
| 1 |
+
{"hellaswag": true, "attention_kind": "selective", "log_dir": "wider_is_better_0/attention_kindselective_n_heads2_seed1340", "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": "wider_is_better_0", "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": 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": 0.00015, "decay_lr": true, "readout_zero_init": false, "query_zero_init": false, "l1_loss": false, "debugpy": false, "key": "1.5e-4_10240_2_1340", "n_embd": 128}
|
attention_kindselective_n_heads2_seed1340/dataloader_02500.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:8b2ea67f78ff5a7970d0db044ff7ee527b3dc065f295fd30f588df4b44b568d0
|
| 3 |
+
size 964
|
attention_kindselective_n_heads2_seed1340/dataloader_04999.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:3fbc8f5b7d2476463a241dae384ac290c173f75d26d125634e27c49e7ea5399c
|
| 3 |
+
size 964
|
attention_kindselective_n_heads2_seed1340/log2.txt
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
attention_kindselective_n_heads2_seed1340/model_02500.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:35bf7922fb61d552001763039749c28ce9e91f1c804595b4ed12461463848071
|
| 3 |
+
size 38587970
|
attention_kindselective_n_heads2_seed1340/model_04999.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:8dcf899c74fa226c12eeb5398c9e0bfaad8680f56de32cdf984df14bed3a5884
|
| 3 |
+
size 38587970
|
attention_kindselective_n_heads2_seed1340/optimizer_02500.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:af807c8aa36ff5a8eb274063f76a52e6108fbcf1dc954f73d4b4e0fb7cd67506
|
| 3 |
+
size 70895430
|
attention_kindselective_n_heads2_seed1340/optimizer_04999.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:e516bc6c30c8e4dccf6dc405b9d817e92f43d34a62eb89dae0df8beb2b8fa5a1
|
| 3 |
+
size 70895430
|