Upload folder using huggingface_hub
Browse files
400M-Slimpajama-400M-OpenR1-Math-220k/L1R/config.json
ADDED
|
@@ -0,0 +1,19 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"sae_type": "sae",
|
| 3 |
+
"hook_point_in": "blocks.1.hook_resid_post",
|
| 4 |
+
"hook_point_out": "blocks.1.hook_resid_post",
|
| 5 |
+
"d_model": 4096,
|
| 6 |
+
"expansion_factor": 8,
|
| 7 |
+
"use_decoder_bias": true,
|
| 8 |
+
"use_glu_encoder": false,
|
| 9 |
+
"act_fn": "jumprelu",
|
| 10 |
+
"jump_relu_threshold": 0.0,
|
| 11 |
+
"apply_decoder_bias_to_pre_encoder": false,
|
| 12 |
+
"norm_activation": "dataset-wise",
|
| 13 |
+
"sparsity_include_decoder_norm": false,
|
| 14 |
+
"force_unit_decoder_norm": false,
|
| 15 |
+
"top_k": 50,
|
| 16 |
+
"use_triton_kernel": true,
|
| 17 |
+
"sparsity_threshold_for_triton_spmm_kernel": 0.99,
|
| 18 |
+
"jumprelu_threshold_window": 2.0
|
| 19 |
+
}
|
400M-Slimpajama-400M-OpenR1-Math-220k/L1R/sae_weights.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:61aa28f60e708f598adf64c546885808a50bcb5a5d9b81e798fbf207818fc1e6
|
| 3 |
+
size 1074020940
|