Update config.json
Browse files- config.json +1 -108
config.json
CHANGED
|
@@ -3,96 +3,6 @@
|
|
| 3 |
"Qwen2ForCausalLM"
|
| 4 |
],
|
| 5 |
"attention_dropout": 0.0,
|
| 6 |
-
"attention_partition_spec": [
|
| 7 |
-
[
|
| 8 |
-
"dp",
|
| 9 |
-
"fsdp"
|
| 10 |
-
],
|
| 11 |
-
"sp",
|
| 12 |
-
"tp",
|
| 13 |
-
null
|
| 14 |
-
],
|
| 15 |
-
"attn_mechanism": "vanilla",
|
| 16 |
-
"axis_dims": [
|
| 17 |
-
1,
|
| 18 |
-
-1,
|
| 19 |
-
1,
|
| 20 |
-
1
|
| 21 |
-
],
|
| 22 |
-
"axis_names": [
|
| 23 |
-
"dp",
|
| 24 |
-
"fsdp",
|
| 25 |
-
"tp",
|
| 26 |
-
"sp"
|
| 27 |
-
],
|
| 28 |
-
"backend": null,
|
| 29 |
-
"bias_partition_spec": [
|
| 30 |
-
[
|
| 31 |
-
"dp",
|
| 32 |
-
"fsdp"
|
| 33 |
-
],
|
| 34 |
-
null,
|
| 35 |
-
null,
|
| 36 |
-
null
|
| 37 |
-
],
|
| 38 |
-
"bits": null,
|
| 39 |
-
"block_b": 1,
|
| 40 |
-
"block_k": 128,
|
| 41 |
-
"block_k_dkv": 128,
|
| 42 |
-
"block_k_dq": 128,
|
| 43 |
-
"block_k_major": 128,
|
| 44 |
-
"block_k_major_dkv": 128,
|
| 45 |
-
"block_k_major_dq": 128,
|
| 46 |
-
"block_q": 128,
|
| 47 |
-
"block_q_dkv": 128,
|
| 48 |
-
"block_q_dq": 128,
|
| 49 |
-
"block_q_major_dkv": 128,
|
| 50 |
-
"bos_token_id": 151643,
|
| 51 |
-
"easy_method": "train",
|
| 52 |
-
"embd_pdrop": 0.0,
|
| 53 |
-
"eos_token_id": 151645,
|
| 54 |
-
"fcm_max_ratio": 0.0,
|
| 55 |
-
"fcm_min_ratio": 0.0,
|
| 56 |
-
"generation_attention_partition_spec": [
|
| 57 |
-
[
|
| 58 |
-
"dp",
|
| 59 |
-
"fsdp"
|
| 60 |
-
],
|
| 61 |
-
null,
|
| 62 |
-
"tp",
|
| 63 |
-
null
|
| 64 |
-
],
|
| 65 |
-
"generation_bias_partition_spec": [
|
| 66 |
-
[
|
| 67 |
-
"dp",
|
| 68 |
-
"fsdp"
|
| 69 |
-
],
|
| 70 |
-
null,
|
| 71 |
-
null,
|
| 72 |
-
null
|
| 73 |
-
],
|
| 74 |
-
"generation_query_partition_spec": [
|
| 75 |
-
[
|
| 76 |
-
"dp",
|
| 77 |
-
"fsdp"
|
| 78 |
-
],
|
| 79 |
-
"sp",
|
| 80 |
-
null,
|
| 81 |
-
null
|
| 82 |
-
],
|
| 83 |
-
"hidden_act": "silu",
|
| 84 |
-
"hidden_size": 5120,
|
| 85 |
-
"initializer_range": 0.02,
|
| 86 |
-
"intermediate_size": 13696,
|
| 87 |
-
"key_partition_spec": [
|
| 88 |
-
[
|
| 89 |
-
"dp",
|
| 90 |
-
"fsdp"
|
| 91 |
-
],
|
| 92 |
-
"sp",
|
| 93 |
-
"tp",
|
| 94 |
-
null
|
| 95 |
-
],
|
| 96 |
"max_position_embeddings": 32768,
|
| 97 |
"max_window_layers": 35,
|
| 98 |
"model_type": "qwen2",
|
|
@@ -100,15 +10,7 @@
|
|
| 100 |
"num_hidden_layers": 40,
|
| 101 |
"num_key_value_heads": 40,
|
| 102 |
"number_rep_kv": 1,
|
| 103 |
-
|
| 104 |
-
[
|
| 105 |
-
"dp",
|
| 106 |
-
"fsdp"
|
| 107 |
-
],
|
| 108 |
-
"sp",
|
| 109 |
-
"tp",
|
| 110 |
-
null
|
| 111 |
-
],
|
| 112 |
"resid_pdrop": 0.0,
|
| 113 |
"rms_norm_eps": 1e-06,
|
| 114 |
"rope_scaling": null,
|
|
@@ -127,14 +29,5 @@
|
|
| 127 |
"use_sharded_kv_caching": true,
|
| 128 |
"use_sharding_constraint": false,
|
| 129 |
"use_sliding_window": false,
|
| 130 |
-
"value_partition_spec": [
|
| 131 |
-
[
|
| 132 |
-
"dp",
|
| 133 |
-
"fsdp"
|
| 134 |
-
],
|
| 135 |
-
"sp",
|
| 136 |
-
"tp",
|
| 137 |
-
null
|
| 138 |
-
],
|
| 139 |
"vocab_size": 152064
|
| 140 |
}
|
|
|
|
| 3 |
"Qwen2ForCausalLM"
|
| 4 |
],
|
| 5 |
"attention_dropout": 0.0,
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 6 |
"max_position_embeddings": 32768,
|
| 7 |
"max_window_layers": 35,
|
| 8 |
"model_type": "qwen2",
|
|
|
|
| 10 |
"num_hidden_layers": 40,
|
| 11 |
"num_key_value_heads": 40,
|
| 12 |
"number_rep_kv": 1,
|
| 13 |
+
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 14 |
"resid_pdrop": 0.0,
|
| 15 |
"rms_norm_eps": 1e-06,
|
| 16 |
"rope_scaling": null,
|
|
|
|
| 29 |
"use_sharded_kv_caching": true,
|
| 30 |
"use_sharding_constraint": false,
|
| 31 |
"use_sliding_window": false,
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 32 |
"vocab_size": 152064
|
| 33 |
}
|