Rename folders
Browse files
Llama3_1Base-L28A-32x/checkpoints/final.safetensors
DELETED
@@ -1,3 +0,0 @@
|
|
1 |
-
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:f53525cdc32e38a8d3e24c0e62e8c802a4197bc65ed506c0c47b1cf275f0ca7d
|
3 |
-
size 2147754376
|
|
|
|
|
|
|
|
Llama3_1Base-L28A-32x/hyperparams.json
DELETED
@@ -1,35 +0,0 @@
|
|
1 |
-
{
|
2 |
-
"device": "cuda:0",
|
3 |
-
"seed": 42,
|
4 |
-
"dtype": "torch.bfloat16",
|
5 |
-
"hook_point_in": "blocks.28.hook_attn_out",
|
6 |
-
"hook_point_out": "blocks.28.hook_attn_out",
|
7 |
-
"use_decoder_bias": true,
|
8 |
-
"apply_decoder_bias_to_pre_encoder": false,
|
9 |
-
"expansion_factor": 32,
|
10 |
-
"d_model": 4096,
|
11 |
-
"d_sae": 131072,
|
12 |
-
"bias_init_method": "all_zero",
|
13 |
-
"act_fn": "jumprelu",
|
14 |
-
"jump_relu_threshold": 0.20703125,
|
15 |
-
"norm_activation": "dataset-wise",
|
16 |
-
"dataset_average_activation_norm": {
|
17 |
-
"in": 5.84375,
|
18 |
-
"out": 5.84375
|
19 |
-
},
|
20 |
-
"decoder_exactly_fixed_norm": false,
|
21 |
-
"sparsity_include_decoder_norm": true,
|
22 |
-
"use_glu_encoder": false,
|
23 |
-
"init_decoder_norm": 0.5,
|
24 |
-
"init_encoder_norm": null,
|
25 |
-
"init_encoder_with_decoder_transpose": true,
|
26 |
-
"lp": 1,
|
27 |
-
"l1_coefficient": 8e-05,
|
28 |
-
"l1_coefficient_warmup_steps": 78125,
|
29 |
-
"top_k": 50,
|
30 |
-
"k_warmup_steps": 78125,
|
31 |
-
"use_batch_norm_mse": true,
|
32 |
-
"use_ghost_grads": false,
|
33 |
-
"tp_size": 1,
|
34 |
-
"ddp_size": 1
|
35 |
-
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
Llama3_1Base-L28A-32x/lm_config.json
DELETED
@@ -1,10 +0,0 @@
|
|
1 |
-
{
|
2 |
-
"device": "cuda",
|
3 |
-
"seed": 42,
|
4 |
-
"dtype": "torch.bfloat16",
|
5 |
-
"model_name": "meta-llama/Meta-Llama-3.1-8B",
|
6 |
-
"use_flash_attn": false,
|
7 |
-
"cache_dir": null,
|
8 |
-
"d_model": 4096,
|
9 |
-
"local_files_only": false
|
10 |
-
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|