ekdeepslubana commited on
Commit
aaf69b2
·
verified ·
1 Parent(s): ef989ec

Upload folder using huggingface_hub

Browse files
sparsemax_dist_corr_0.2_seed_1134/conf.yaml ADDED
@@ -0,0 +1,40 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ bf16: false
2
+ data:
3
+ batch_size: 64
4
+ corr_config: ../configs/corr_ds-sp_0.2.jsonl
5
+ dev_path: ../data/labeled_sentences_large_deduped_dev.jsonl
6
+ max_sample_length: 64
7
+ num_iters: 10000
8
+ num_workers: 2
9
+ path: ../data/labeled_sentences_large_deduped_train.jsonl
10
+ run_validation: false
11
+ deploy: true
12
+ device: cuda:0
13
+ epochs: 1
14
+ eval:
15
+ save_tables: false
16
+ log:
17
+ eval_interval: 1000
18
+ log_interval: 10
19
+ save_interval: 1000
20
+ save_multiple: false
21
+ wandb_project_name: entangled_saes
22
+ model_name: google/gemma-2-2b
23
+ optimizer:
24
+ beta1: 0.9
25
+ beta2: 0.95
26
+ decay_lr: true
27
+ grad_clip: 1.0
28
+ learning_rate: 0.001
29
+ min_lr: 0.0009
30
+ warmup_iters: 200
31
+ weight_decay: 0.0001
32
+ sae:
33
+ encoder_reg: true
34
+ exp_factor: 8
35
+ gamma_reg: 10
36
+ kval_topk: 128
37
+ mp_kval: 128
38
+ sae_type: sparsemax_dist
39
+ seed: 1134
40
+ tag: sparsemax_dist_corr_0.2_seed_1134
sparsemax_dist_corr_0.2_seed_1134/latest_ckpt.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:693cd44a264bfca287d5212568e981ab56fd1c0a0b232e6524025926ae50d895
3
+ size 1019308980