Commit
•
d787118
1
Parent(s):
73efe29
Upload config.json with huggingface_hub
Browse files- config.json +233 -0
config.json
ADDED
|
@@ -0,0 +1,233 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"architectures": [
|
| 3 |
+
"JambaForCausalLM"
|
| 4 |
+
],
|
| 5 |
+
"attention_dropout": 0.0,
|
| 6 |
+
"attn_hidden_size": -1,
|
| 7 |
+
"attn_implementation": "flex",
|
| 8 |
+
"attn_layer_offset": 100,
|
| 9 |
+
"attn_layer_period": 1,
|
| 10 |
+
"attn_reuse_every_i_layer": -1,
|
| 11 |
+
"bos_token_id": 1,
|
| 12 |
+
"calc_logits_for_entire_prompt": false,
|
| 13 |
+
"compact_gating": false,
|
| 14 |
+
"compute_attn_mat": false,
|
| 15 |
+
"conv_dim": {
|
| 16 |
+
"0": 3200,
|
| 17 |
+
"1": 3200,
|
| 18 |
+
"2": 3200,
|
| 19 |
+
"3": 3200,
|
| 20 |
+
"4": 3200,
|
| 21 |
+
"5": 3200,
|
| 22 |
+
"6": 3200,
|
| 23 |
+
"7": 3200,
|
| 24 |
+
"8": 3200,
|
| 25 |
+
"9": 3200,
|
| 26 |
+
"10": 3200,
|
| 27 |
+
"11": 3200,
|
| 28 |
+
"12": 3200,
|
| 29 |
+
"13": 3200,
|
| 30 |
+
"14": 3200,
|
| 31 |
+
"15": 3200,
|
| 32 |
+
"16": 3200,
|
| 33 |
+
"17": 3200,
|
| 34 |
+
"18": 3200,
|
| 35 |
+
"19": 3200,
|
| 36 |
+
"20": 3200,
|
| 37 |
+
"21": 3200,
|
| 38 |
+
"22": 3200,
|
| 39 |
+
"23": 3200,
|
| 40 |
+
"24": 3200,
|
| 41 |
+
"25": 3200,
|
| 42 |
+
"26": 3200,
|
| 43 |
+
"27": 3200,
|
| 44 |
+
"28": 3200,
|
| 45 |
+
"29": 3200,
|
| 46 |
+
"30": 3200,
|
| 47 |
+
"31": 3200
|
| 48 |
+
},
|
| 49 |
+
"dense_public_ffn_structure": false,
|
| 50 |
+
"double_v_dim": false,
|
| 51 |
+
"enable_mod": false,
|
| 52 |
+
"eos_token_id": 2,
|
| 53 |
+
"expert_layer_offset": 1000,
|
| 54 |
+
"expert_layer_period": 1000,
|
| 55 |
+
"ffn_reuse_every_i_layer": -1,
|
| 56 |
+
"ffn_sharing_config": null,
|
| 57 |
+
"fully_parallel_jamba": false,
|
| 58 |
+
"fused_multihead_config": {
|
| 59 |
+
"conv_attn": null,
|
| 60 |
+
"diverse_head_merge_op": "mean_wo_gate",
|
| 61 |
+
"expand_v": true,
|
| 62 |
+
"qkv_mode": "orig",
|
| 63 |
+
"share_qk": false
|
| 64 |
+
},
|
| 65 |
+
"global_attn_idx": [
|
| 66 |
+
0,
|
| 67 |
+
15,
|
| 68 |
+
31
|
| 69 |
+
],
|
| 70 |
+
"gradient_checkpoint_layer": null,
|
| 71 |
+
"hidden_act": "silu",
|
| 72 |
+
"hidden_size": 1600,
|
| 73 |
+
"hybrid_block_indices": [],
|
| 74 |
+
"hybrid_decoder_layer": "mamba",
|
| 75 |
+
"initializer_range": 0.02,
|
| 76 |
+
"intermediate_size": 5504,
|
| 77 |
+
"kq_head_dim": -1,
|
| 78 |
+
"kq_norm": "none",
|
| 79 |
+
"kv_reuse_every_i_layer": -1,
|
| 80 |
+
"kv_reuse_group": [
|
| 81 |
+
[
|
| 82 |
+
1,
|
| 83 |
+
2
|
| 84 |
+
],
|
| 85 |
+
[
|
| 86 |
+
3,
|
| 87 |
+
4
|
| 88 |
+
],
|
| 89 |
+
[
|
| 90 |
+
5,
|
| 91 |
+
6
|
| 92 |
+
],
|
| 93 |
+
[
|
| 94 |
+
7,
|
| 95 |
+
8
|
| 96 |
+
],
|
| 97 |
+
[
|
| 98 |
+
9,
|
| 99 |
+
10
|
| 100 |
+
],
|
| 101 |
+
[
|
| 102 |
+
11,
|
| 103 |
+
12
|
| 104 |
+
],
|
| 105 |
+
[
|
| 106 |
+
13,
|
| 107 |
+
14
|
| 108 |
+
],
|
| 109 |
+
[
|
| 110 |
+
16,
|
| 111 |
+
17,
|
| 112 |
+
18
|
| 113 |
+
],
|
| 114 |
+
[
|
| 115 |
+
19,
|
| 116 |
+
20
|
| 117 |
+
],
|
| 118 |
+
[
|
| 119 |
+
21,
|
| 120 |
+
22
|
| 121 |
+
],
|
| 122 |
+
[
|
| 123 |
+
23,
|
| 124 |
+
24
|
| 125 |
+
],
|
| 126 |
+
[
|
| 127 |
+
25,
|
| 128 |
+
26
|
| 129 |
+
],
|
| 130 |
+
[
|
| 131 |
+
27,
|
| 132 |
+
28
|
| 133 |
+
],
|
| 134 |
+
[
|
| 135 |
+
29,
|
| 136 |
+
30
|
| 137 |
+
]
|
| 138 |
+
],
|
| 139 |
+
"kv_weight_reuse": false,
|
| 140 |
+
"layer_type": [
|
| 141 |
+
"h",
|
| 142 |
+
"h",
|
| 143 |
+
"h",
|
| 144 |
+
"h",
|
| 145 |
+
"h",
|
| 146 |
+
"h",
|
| 147 |
+
"h",
|
| 148 |
+
"h",
|
| 149 |
+
"h",
|
| 150 |
+
"h",
|
| 151 |
+
"h",
|
| 152 |
+
"h",
|
| 153 |
+
"h",
|
| 154 |
+
"h",
|
| 155 |
+
"h",
|
| 156 |
+
"h",
|
| 157 |
+
"h",
|
| 158 |
+
"h",
|
| 159 |
+
"h",
|
| 160 |
+
"h",
|
| 161 |
+
"h",
|
| 162 |
+
"h",
|
| 163 |
+
"h",
|
| 164 |
+
"h",
|
| 165 |
+
"h",
|
| 166 |
+
"h",
|
| 167 |
+
"h",
|
| 168 |
+
"h",
|
| 169 |
+
"h",
|
| 170 |
+
"h",
|
| 171 |
+
"h",
|
| 172 |
+
"h"
|
| 173 |
+
],
|
| 174 |
+
"local_expand_ratio": 1,
|
| 175 |
+
"local_global_dual_branch": false,
|
| 176 |
+
"local_global_dual_branch_merge_op": "mean",
|
| 177 |
+
"lookback_mode": "",
|
| 178 |
+
"macro_arch": "",
|
| 179 |
+
"mamba2_headdim": 64,
|
| 180 |
+
"mamba_conv_bias": true,
|
| 181 |
+
"mamba_d_conv": 4,
|
| 182 |
+
"mamba_d_state": 16,
|
| 183 |
+
"mamba_dt_rank": 100,
|
| 184 |
+
"mamba_expand": 2,
|
| 185 |
+
"mamba_inner_layernorms": true,
|
| 186 |
+
"mamba_latent_size": null,
|
| 187 |
+
"mamba_multihead_config": null,
|
| 188 |
+
"mamba_proj_bias": false,
|
| 189 |
+
"mamba_reuse_every_i_layer": -1,
|
| 190 |
+
"max_position_embeddings": 2176,
|
| 191 |
+
"memory_tokens_interspersed_every": 0,
|
| 192 |
+
"mlp_hidden_act": "silu",
|
| 193 |
+
"mod_topk": 2,
|
| 194 |
+
"model_type": "jamba",
|
| 195 |
+
"moe_config": null,
|
| 196 |
+
"num_attention_heads": 25,
|
| 197 |
+
"num_attn_per_ffn": 3,
|
| 198 |
+
"num_experts": 1,
|
| 199 |
+
"num_experts_per_tok": 1,
|
| 200 |
+
"num_ffn": 1,
|
| 201 |
+
"num_hidden_layers": 32,
|
| 202 |
+
"num_key_value_heads": 5,
|
| 203 |
+
"num_mamba": 1,
|
| 204 |
+
"num_memory_tokens": 128,
|
| 205 |
+
"output_router_logits": false,
|
| 206 |
+
"pad_token_id": 0,
|
| 207 |
+
"public_ffn_structure": false,
|
| 208 |
+
"reduce_attn_ratio": 0.5,
|
| 209 |
+
"reduce_method": "mean",
|
| 210 |
+
"rms_norm_eps": 1e-06,
|
| 211 |
+
"rope": true,
|
| 212 |
+
"rope_theta": 10000.0,
|
| 213 |
+
"rope_type": null,
|
| 214 |
+
"router_aux_loss_coef": 0.001,
|
| 215 |
+
"save_input_output": false,
|
| 216 |
+
"seq_length": 2048,
|
| 217 |
+
"sequential_jamba": false,
|
| 218 |
+
"share_kv": false,
|
| 219 |
+
"shared_module_attn": "",
|
| 220 |
+
"shared_module_mamba": "",
|
| 221 |
+
"sliding_window": 1024,
|
| 222 |
+
"sliding_window_size": null,
|
| 223 |
+
"swa_full_head": false,
|
| 224 |
+
"tie_word_embeddings": true,
|
| 225 |
+
"torch_dtype": "bfloat16",
|
| 226 |
+
"transformers_version": "4.44.1",
|
| 227 |
+
"use_cache": false,
|
| 228 |
+
"use_mamba2": false,
|
| 229 |
+
"use_mamba_kernels": true,
|
| 230 |
+
"v_head_dim": 128,
|
| 231 |
+
"visual_attn": false,
|
| 232 |
+
"vocab_size": 32001
|
| 233 |
+
}
|