Upload config.json
Browse files- config.json +30 -0
config.json
ADDED
|
@@ -0,0 +1,30 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"model_type": "phogpt",
|
| 3 |
+
"architectures": ["PhoGPTForCausalLM"],
|
| 4 |
+
"hidden_size": 4096,
|
| 5 |
+
"num_attention_heads": 32,
|
| 6 |
+
"num_hidden_layers": 32,
|
| 7 |
+
"vocab_size": 51200,
|
| 8 |
+
"max_seq_len": 4096,
|
| 9 |
+
"torch_dtype": "float16",
|
| 10 |
+
"attn_config": {
|
| 11 |
+
"attn_impl": "triton",
|
| 12 |
+
"attn_pdrop": 0.0,
|
| 13 |
+
"attn_type": "multihead_attention",
|
| 14 |
+
"alibi": true,
|
| 15 |
+
"alibi_bias_max": 8,
|
| 16 |
+
"prefix_lm": false,
|
| 17 |
+
"qk_ln": false
|
| 18 |
+
},
|
| 19 |
+
"ffn_config": {
|
| 20 |
+
"ffn_type": "phogpt_mlp"
|
| 21 |
+
},
|
| 22 |
+
"init_config": {
|
| 23 |
+
"name": "kaiming_normal_",
|
| 24 |
+
"fan_mode": "fan_in",
|
| 25 |
+
"init_nonlinearity": "relu",
|
| 26 |
+
"init_div_is_residual": true
|
| 27 |
+
},
|
| 28 |
+
"use_cache": false
|
| 29 |
+
}
|
| 30 |
+
|