Upload beta1_0.0_20240308221813946612_config.json with huggingface_hub
Browse files
beta1_0.0_20240308221813946612_config.json
ADDED
@@ -0,0 +1,32 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"seed": 42,
|
3 |
+
"model_name_or_path": "EleutherAI/pythia-70m-deduped",
|
4 |
+
"hook_point": "gpt_neox.layers.3",
|
5 |
+
"dataset_name_or_path": "jbrinkma/pile-300k",
|
6 |
+
"activation_size": -1,
|
7 |
+
"add_bos_token": false,
|
8 |
+
"expansion_factor": 4,
|
9 |
+
"b_dec_init_method": "",
|
10 |
+
"n_steps": -1,
|
11 |
+
"device": "cuda",
|
12 |
+
"batch_size": 32,
|
13 |
+
"ctx_length": 256,
|
14 |
+
"lr": 0.001,
|
15 |
+
"min_lr": 0.0,
|
16 |
+
"lr_warmup_steps": 5000,
|
17 |
+
"sparsity_coefficient": 0.003,
|
18 |
+
"evaluation_interval": 200,
|
19 |
+
"beta1": 0.0,
|
20 |
+
"beta2": 0.999,
|
21 |
+
"l1_sqrt": false,
|
22 |
+
"n_tokens_in_feature_cache": 500000.0,
|
23 |
+
"use_ghost_grads": false,
|
24 |
+
"output_dir": "outputs",
|
25 |
+
"cache_dir": "cache",
|
26 |
+
"checkpoint_interval": 200,
|
27 |
+
"use_wandb": true,
|
28 |
+
"wandb_entity": "best_sae",
|
29 |
+
"wandb_project": "best_sae",
|
30 |
+
"wandb_name": "beta1_0.0_20240308221813946612",
|
31 |
+
"wandb_group": "Beta1_v1"
|
32 |
+
}
|