Update config.json
Browse files- config.json +24 -5
config.json
CHANGED
@@ -1,8 +1,27 @@
|
|
1 |
{
|
2 |
-
"
|
3 |
-
|
4 |
-
|
5 |
-
"
|
6 |
-
"
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
7 |
"transformers_version": "4.44.0"
|
8 |
}
|
|
|
1 |
{
|
2 |
+
"model_type": "causal_lm",
|
3 |
+
"base_model_name_or_path": "luxmorocco/dusturia-3.1",
|
4 |
+
"max_position_embeddings": 2048,
|
5 |
+
"hidden_size": 8192,
|
6 |
+
"num_attention_heads": 64,
|
7 |
+
"num_hidden_layers": 32,
|
8 |
+
"intermediate_size": 32768,
|
9 |
+
"hidden_act": "gelu",
|
10 |
+
"initializer_range": 0.02,
|
11 |
+
"layer_norm_eps": 1e-6,
|
12 |
+
"vocab_size": 50265,
|
13 |
+
"bos_token_id": 0,
|
14 |
+
"eos_token_id": 1,
|
15 |
+
"pad_token_id": 2,
|
16 |
+
"gradient_checkpointing": true,
|
17 |
+
"task_specific_params": {
|
18 |
+
"text-generation": {
|
19 |
+
"do_sample": true,
|
20 |
+
"max_length": 200,
|
21 |
+
"temperature": 0.7,
|
22 |
+
"top_p": 0.95,
|
23 |
+
"top_k": 50
|
24 |
+
}
|
25 |
+
},
|
26 |
"transformers_version": "4.44.0"
|
27 |
}
|