model upload
Browse files- config.json +22 -1
config.json
CHANGED
|
@@ -6,5 +6,26 @@
|
|
| 6 |
"task": "question-answering",
|
| 7 |
"language": "th",
|
| 8 |
"max_sequence_length": 2048,
|
| 9 |
-
"do_sample": true
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 10 |
}
|
|
|
|
| 6 |
"task": "question-answering",
|
| 7 |
"language": "th",
|
| 8 |
"max_sequence_length": 2048,
|
| 9 |
+
"do_sample": true,
|
| 10 |
+
"attention_bias": false,
|
| 11 |
+
"attention_dropout": 0.0,
|
| 12 |
+
"bos_token_id": 128000,
|
| 13 |
+
"eos_token_id": 128001,
|
| 14 |
+
"hidden_act": "silu",
|
| 15 |
+
"hidden_size": 4096,
|
| 16 |
+
"initializer_range": 0.02,
|
| 17 |
+
"intermediate_size": 14336,
|
| 18 |
+
"max_position_embeddings": 8192,
|
| 19 |
+
"num_attention_heads": 32,
|
| 20 |
+
"num_hidden_layers": 32,
|
| 21 |
+
"num_key_value_heads": 8,
|
| 22 |
+
"pretraining_tp": 1,
|
| 23 |
+
"rms_norm_eps": 1e-05,
|
| 24 |
+
"rope_scaling": null,
|
| 25 |
+
"rope_theta": 500000.0,
|
| 26 |
+
"tie_word_embeddings": false,
|
| 27 |
+
"torch_dtype": "bfloat16",
|
| 28 |
+
"transformers_version": "4.40.0",
|
| 29 |
+
"use_cache": true,
|
| 30 |
+
"vocab_size": 128256
|
| 31 |
}
|