Upload config.json with huggingface_hub
Browse files- config.json +115 -0
config.json
ADDED
|
@@ -0,0 +1,115 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"activation_dims": {
|
| 3 |
+
"mlp_0": 768,
|
| 4 |
+
"attn_0": 768,
|
| 5 |
+
"mlp_1": 768,
|
| 6 |
+
"attn_1": 768,
|
| 7 |
+
"mlp_2": 768,
|
| 8 |
+
"attn_2": 768,
|
| 9 |
+
"mlp_3": 768,
|
| 10 |
+
"attn_3": 768,
|
| 11 |
+
"mlp_4": 768,
|
| 12 |
+
"attn_4": 768,
|
| 13 |
+
"mlp_5": 768,
|
| 14 |
+
"attn_5": 768,
|
| 15 |
+
"mlp_6": 768,
|
| 16 |
+
"attn_6": 768,
|
| 17 |
+
"mlp_7": 768,
|
| 18 |
+
"attn_7": 768,
|
| 19 |
+
"mlp_8": 768,
|
| 20 |
+
"attn_8": 768,
|
| 21 |
+
"mlp_9": 768,
|
| 22 |
+
"attn_9": 768,
|
| 23 |
+
"mlp_10": 768,
|
| 24 |
+
"attn_10": 768,
|
| 25 |
+
"mlp_11": 768,
|
| 26 |
+
"attn_11": 768
|
| 27 |
+
},
|
| 28 |
+
"dict_sizes": {
|
| 29 |
+
"mlp_0": 6144,
|
| 30 |
+
"attn_0": 6144,
|
| 31 |
+
"mlp_1": 6144,
|
| 32 |
+
"attn_1": 6144,
|
| 33 |
+
"mlp_2": 6144,
|
| 34 |
+
"attn_2": 6144,
|
| 35 |
+
"mlp_3": 6144,
|
| 36 |
+
"attn_3": 6144,
|
| 37 |
+
"mlp_4": 6144,
|
| 38 |
+
"attn_4": 6144,
|
| 39 |
+
"mlp_5": 6144,
|
| 40 |
+
"attn_5": 6144,
|
| 41 |
+
"mlp_6": 6144,
|
| 42 |
+
"attn_6": 6144,
|
| 43 |
+
"mlp_7": 6144,
|
| 44 |
+
"attn_7": 6144,
|
| 45 |
+
"mlp_8": 6144,
|
| 46 |
+
"attn_8": 6144,
|
| 47 |
+
"mlp_9": 6144,
|
| 48 |
+
"attn_9": 6144,
|
| 49 |
+
"mlp_10": 6144,
|
| 50 |
+
"attn_10": 6144,
|
| 51 |
+
"mlp_11": 6144,
|
| 52 |
+
"attn_11": 6144
|
| 53 |
+
},
|
| 54 |
+
"ks": {
|
| 55 |
+
"mlp_0": 64,
|
| 56 |
+
"attn_0": 64,
|
| 57 |
+
"mlp_1": 64,
|
| 58 |
+
"attn_1": 64,
|
| 59 |
+
"mlp_2": 64,
|
| 60 |
+
"attn_2": 64,
|
| 61 |
+
"mlp_3": 64,
|
| 62 |
+
"attn_3": 64,
|
| 63 |
+
"mlp_4": 64,
|
| 64 |
+
"attn_4": 64,
|
| 65 |
+
"mlp_5": 64,
|
| 66 |
+
"attn_5": 64,
|
| 67 |
+
"mlp_6": 64,
|
| 68 |
+
"attn_6": 64,
|
| 69 |
+
"mlp_7": 64,
|
| 70 |
+
"attn_7": 64,
|
| 71 |
+
"mlp_8": 64,
|
| 72 |
+
"attn_8": 64,
|
| 73 |
+
"mlp_9": 64,
|
| 74 |
+
"attn_9": 64,
|
| 75 |
+
"mlp_10": 64,
|
| 76 |
+
"attn_10": 64,
|
| 77 |
+
"mlp_11": 64,
|
| 78 |
+
"attn_11": 64
|
| 79 |
+
},
|
| 80 |
+
"layers": [],
|
| 81 |
+
"lm_name": "",
|
| 82 |
+
"submodule_names": [
|
| 83 |
+
"mlp_0",
|
| 84 |
+
"attn_0",
|
| 85 |
+
"mlp_1",
|
| 86 |
+
"attn_1",
|
| 87 |
+
"mlp_2",
|
| 88 |
+
"attn_2",
|
| 89 |
+
"mlp_3",
|
| 90 |
+
"attn_3",
|
| 91 |
+
"mlp_4",
|
| 92 |
+
"attn_4",
|
| 93 |
+
"mlp_5",
|
| 94 |
+
"attn_5",
|
| 95 |
+
"mlp_6",
|
| 96 |
+
"attn_6",
|
| 97 |
+
"mlp_7",
|
| 98 |
+
"attn_7",
|
| 99 |
+
"mlp_8",
|
| 100 |
+
"attn_8",
|
| 101 |
+
"mlp_9",
|
| 102 |
+
"attn_9",
|
| 103 |
+
"mlp_10",
|
| 104 |
+
"attn_10",
|
| 105 |
+
"mlp_11",
|
| 106 |
+
"attn_11"
|
| 107 |
+
],
|
| 108 |
+
"connection_sparsity_coeff": 0.01,
|
| 109 |
+
"use_sparse_connections": false,
|
| 110 |
+
"buffer_config": {
|
| 111 |
+
"ctx_len": 128,
|
| 112 |
+
"refresh_batch_size": 256,
|
| 113 |
+
"out_batch_size": 4096
|
| 114 |
+
}
|
| 115 |
+
}
|