Upload Config.json
Browse files- config.json +27 -0
config.json
ADDED
|
@@ -0,0 +1,27 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"architectures": [
|
| 3 |
+
"YENForCausalLM"
|
| 4 |
+
],
|
| 5 |
+
"model_type": "yen",
|
| 6 |
+
"torch_dtype": "float32",
|
| 7 |
+
"transformers_version": "4.49.0.dev0",
|
| 8 |
+
"vocab_size": 256,
|
| 9 |
+
"hidden_size": 1,
|
| 10 |
+
"num_hidden_layers": 1,
|
| 11 |
+
"num_attention_heads": 1,
|
| 12 |
+
"layer_norm_eps": 1e-06,
|
| 13 |
+
"max_position_embeddings": 15000000,
|
| 14 |
+
"rope_theta": 10000000.0,
|
| 15 |
+
"rope_scaling": {
|
| 16 |
+
"type": "dynamic",
|
| 17 |
+
"factor": 2.0
|
| 18 |
+
},
|
| 19 |
+
"activation_function": "gelu",
|
| 20 |
+
"tie_word_embeddings": false,
|
| 21 |
+
"use_cache": true,
|
| 22 |
+
"intermediate_size": 512,
|
| 23 |
+
"classifier_dropout": 0.1,
|
| 24 |
+
"attention_dropout": 0.1,
|
| 25 |
+
"bos_token_id": 0,
|
| 26 |
+
"eos_token_id": 255
|
| 27 |
+
}
|