Upload config.json with huggingface_hub
Browse files- config.json +50 -0
config.json
ADDED
|
@@ -0,0 +1,50 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"apply_residual_connection_post_layernorm": false,
|
| 3 |
+
"architectures": [
|
| 4 |
+
"TeleChat2ForCausalLM"
|
| 5 |
+
],
|
| 6 |
+
"attention_dropout": 0.0,
|
| 7 |
+
"attention_softmax_in_fp32": true,
|
| 8 |
+
"auto_map": {
|
| 9 |
+
"AutoConfig": "configuration_telechat2.Telechat2Config",
|
| 10 |
+
"AutoModelForCausalLM": "modeling_telechat2.Telechat2ForCausalLM"
|
| 11 |
+
},
|
| 12 |
+
"base_seqlen": 32768,
|
| 13 |
+
"bias_dropout_fusion": true,
|
| 14 |
+
"bos_token_id": 1,
|
| 15 |
+
"embed_layernorm": false,
|
| 16 |
+
"eos_token_id": 2,
|
| 17 |
+
"ffn_hidden_size": 20480,
|
| 18 |
+
"flash_attn": true,
|
| 19 |
+
"hidden_dropout": 0.0,
|
| 20 |
+
"hidden_size": 6144,
|
| 21 |
+
"initializer_range": 0.02,
|
| 22 |
+
"layer_norm_epsilon": 1e-05,
|
| 23 |
+
"logn": true,
|
| 24 |
+
"masked_softmax_fusion": true,
|
| 25 |
+
"max_position_embeddings": 32768,
|
| 26 |
+
"model_type": "telechat",
|
| 27 |
+
"n_head": 48,
|
| 28 |
+
"n_inner": null,
|
| 29 |
+
"n_layer": 64,
|
| 30 |
+
"num_key_value_heads": 48,
|
| 31 |
+
"offset_alibi": 100,
|
| 32 |
+
"pad_token_id": 3,
|
| 33 |
+
"pretraining_tp": 2,
|
| 34 |
+
"rope_scaling": {
|
| 35 |
+
"factor": 1.0,
|
| 36 |
+
"rope_type": "dynamic"
|
| 37 |
+
},
|
| 38 |
+
"rope_theta": 1000000,
|
| 39 |
+
"seq_length": 32768,
|
| 40 |
+
"skip_bias_add": false,
|
| 41 |
+
"skip_bias_add_qkv": false,
|
| 42 |
+
"slow_but_exact": false,
|
| 43 |
+
"tie_word_embeddings": false,
|
| 44 |
+
"torch_dtype": "bfloat16",
|
| 45 |
+
"training_seqlen": 32768,
|
| 46 |
+
"transformers_version": "4.48.2",
|
| 47 |
+
"unk_token_id": 0,
|
| 48 |
+
"use_cache": false,
|
| 49 |
+
"vocab_size": 131072
|
| 50 |
+
}
|