Hugging Face's logo

andrew-healey
/
context-compression

Model card Files
xet
Community
andrew-healey commited on
Commit
20e75b7
·
verified ·
1 Parent(s): c7f7d5d

Upload folder using huggingface_hub

Browse files
Files changed (8) hide show
  1. lr1e-4_total_batch_size5120_seq_len256_decay_lrfalse/args.json +1 -0
  2. lr1e-4_total_batch_size5120_seq_len256_decay_lrfalse/dataloader_02500.pt +3 -0
  3. lr1e-4_total_batch_size5120_seq_len256_decay_lrfalse/dataloader_04999.pt +3 -0
  4. lr1e-4_total_batch_size5120_seq_len256_decay_lrfalse/log2.txt +0 -0
  5. lr1e-4_total_batch_size5120_seq_len256_decay_lrfalse/model_02500.pt +3 -0
  6. lr1e-4_total_batch_size5120_seq_len256_decay_lrfalse/model_04999.pt +3 -0
  7. lr1e-4_total_batch_size5120_seq_len256_decay_lrfalse/optimizer_02500.pt +3 -0
  8. lr1e-4_total_batch_size5120_seq_len256_decay_lrfalse/optimizer_04999.pt +3 -0
lr1e-4_total_batch_size5120_seq_len256_decay_lrfalse/args.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {"hellaswag": true, "attention_kind": "selective", "log_dir": "proxy_model_sweep_2/lr1e-4_total_batch_size5120_seq_len256_decay_lrfalse", "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": 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": true, "mup_enable_coord_check_logging": false, "max_lr": 0.0001, "decay_lr": false, "readout_zero_init": false, "query_zero_init": false, "l1_loss": false, "debugpy": false, "n_embd": 128}
lr1e-4_total_batch_size5120_seq_len256_decay_lrfalse/dataloader_02500.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f5f86eba64e08aa3145708d463ef689ecaee47ebe0a97035e6e0aeb59f377344
3
+ size 964
lr1e-4_total_batch_size5120_seq_len256_decay_lrfalse/dataloader_04999.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:431842dc67f0d83e5eb85cf9c36fd024e295001fe78f3ba5deeae77865394feb
3
+ size 964
lr1e-4_total_batch_size5120_seq_len256_decay_lrfalse/log2.txt ADDED
The diff for this file is too large to render. See raw diff
 
lr1e-4_total_batch_size5120_seq_len256_decay_lrfalse/model_02500.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:0a5822d54263a85aeb9de9c4350eccbd35c20fbd7edcbf07681881aecfad6040
3
+ size 38587970
lr1e-4_total_batch_size5120_seq_len256_decay_lrfalse/model_04999.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:7af84fcc0fef6f291d5e6b0a482eda2878143d116747d051d3fc6a8ce71c0e7f
3
+ size 38587970
lr1e-4_total_batch_size5120_seq_len256_decay_lrfalse/optimizer_02500.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:7beda8040b0669a2b01b104aa5e952549ba084416b2008168decb074203473fa
3
+ size 70895430
lr1e-4_total_batch_size5120_seq_len256_decay_lrfalse/optimizer_04999.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:1fe26361b1f2791432ef22524cf3f5660dc375273ded754e8dea226ea16e0c4a
3
+ size 70895430