Upload 2 files
Browse files
tied_per_target_skip_global_batchtopk_jumprelu/cfg.json
ADDED
|
@@ -0,0 +1,26 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"num_features": 122880,
|
| 3 |
+
"num_layers": 12,
|
| 4 |
+
"d_model": 768,
|
| 5 |
+
"model_name": null,
|
| 6 |
+
"normalization_method": "mean_std",
|
| 7 |
+
"activation_fn": "jumprelu",
|
| 8 |
+
"jumprelu_threshold": 0.0,
|
| 9 |
+
"batchtopk_k": null,
|
| 10 |
+
"batchtopk_straight_through": false,
|
| 11 |
+
"topk_k": null,
|
| 12 |
+
"topk_straight_through": true,
|
| 13 |
+
"topk_mode": "global",
|
| 14 |
+
"two_stage_batchtopk": false,
|
| 15 |
+
"two_stage_topk": false,
|
| 16 |
+
"clt_dtype": null,
|
| 17 |
+
"expected_input_dtype": null,
|
| 18 |
+
"mlp_input_template": null,
|
| 19 |
+
"mlp_output_template": null,
|
| 20 |
+
"tl_input_template": null,
|
| 21 |
+
"tl_output_template": null,
|
| 22 |
+
"decoder_tying": "per_target",
|
| 23 |
+
"enable_feature_offset": false,
|
| 24 |
+
"enable_feature_scale": false,
|
| 25 |
+
"skip_connection": true
|
| 26 |
+
}
|
tied_per_target_skip_global_batchtopk_jumprelu/tied_per_target_skip_global_batchtopk_jumprelu.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:521375285e2b0c5baf0b3a9ca02f35869913fddc0107c7e02baf532f1b4d4366
|
| 3 |
+
size 9099848184
|