Hugging Face's logo

andrew-healey
/
context-compression

Model card Files
xet
Community
andrew-healey commited on
Commit
80fd823
·
verified ·
1 Parent(s): d30d95d

Upload folder using huggingface_hub

Browse files
Files changed (8) hide show
  1. lr1.5e-4_total_batch_size5120_seq_len256_decay_lrtrue_attention_kindselective_warmup_steps1/args.json +1 -0
  2. lr1.5e-4_total_batch_size5120_seq_len256_decay_lrtrue_attention_kindselective_warmup_steps1/dataloader_02500.pt +3 -0
  3. lr1.5e-4_total_batch_size5120_seq_len256_decay_lrtrue_attention_kindselective_warmup_steps1/dataloader_04999.pt +3 -0
  4. lr1.5e-4_total_batch_size5120_seq_len256_decay_lrtrue_attention_kindselective_warmup_steps1/log2.txt +0 -0
  5. lr1.5e-4_total_batch_size5120_seq_len256_decay_lrtrue_attention_kindselective_warmup_steps1/model_02500.pt +3 -0
  6. lr1.5e-4_total_batch_size5120_seq_len256_decay_lrtrue_attention_kindselective_warmup_steps1/model_04999.pt +3 -0
  7. lr1.5e-4_total_batch_size5120_seq_len256_decay_lrtrue_attention_kindselective_warmup_steps1/optimizer_02500.pt +3 -0
  8. lr1.5e-4_total_batch_size5120_seq_len256_decay_lrtrue_attention_kindselective_warmup_steps1/optimizer_04999.pt +3 -0
lr1.5e-4_total_batch_size5120_seq_len256_decay_lrtrue_attention_kindselective_warmup_steps1/args.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {"hellaswag": true, "attention_kind": "selective", "log_dir": "proxy_model_sweep_2/lr1.5e-4_total_batch_size5120_seq_len256_decay_lrtrue_attention_kindselective_warmup_steps1", "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": 1, "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": 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": 0.00015, "decay_lr": true, "readout_zero_init": false, "query_zero_init": false, "l1_loss": false, "debugpy": false, "n_embd": 128}
lr1.5e-4_total_batch_size5120_seq_len256_decay_lrtrue_attention_kindselective_warmup_steps1/dataloader_02500.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f5f86eba64e08aa3145708d463ef689ecaee47ebe0a97035e6e0aeb59f377344
3
+ size 964
lr1.5e-4_total_batch_size5120_seq_len256_decay_lrtrue_attention_kindselective_warmup_steps1/dataloader_04999.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:431842dc67f0d83e5eb85cf9c36fd024e295001fe78f3ba5deeae77865394feb
3
+ size 964
lr1.5e-4_total_batch_size5120_seq_len256_decay_lrtrue_attention_kindselective_warmup_steps1/log2.txt ADDED
The diff for this file is too large to render. See raw diff
 
lr1.5e-4_total_batch_size5120_seq_len256_decay_lrtrue_attention_kindselective_warmup_steps1/model_02500.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:aeaff4c3f064596a7c6a31d30fa44da60d5ac37739f57a43d6108de1f75eadc8
3
+ size 38587970
lr1.5e-4_total_batch_size5120_seq_len256_decay_lrtrue_attention_kindselective_warmup_steps1/model_04999.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f8be4ca39278c727a550dc8ff9a116d943b26bf4341af9d25029279f1c3411f5
3
+ size 38587970
lr1.5e-4_total_batch_size5120_seq_len256_decay_lrtrue_attention_kindselective_warmup_steps1/optimizer_02500.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:7af69e08de0bb417897064c645b7909aa1b4a744746f3fecb02ea1dfd4f39d25
3
+ size 70895430
lr1.5e-4_total_batch_size5120_seq_len256_decay_lrtrue_attention_kindselective_warmup_steps1/optimizer_04999.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:2fa1a65c6f2c7be7e78ec31be717b793733e13696e46a0e10c30df7a09b8266d
3
+ size 70895430