Upload config.json with huggingface_hub
Browse files- config.json +24 -0
config.json
ADDED
|
@@ -0,0 +1,24 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"model_type": "gptmoe-custom",
|
| 3 |
+
"vocab_size": 32000,
|
| 4 |
+
"d_model": 768,
|
| 5 |
+
"n_heads": 12,
|
| 6 |
+
"n_layers": 16,
|
| 7 |
+
"ffn_mult": 2.6666666666666665,
|
| 8 |
+
"use_rmsnorm": true,
|
| 9 |
+
"rms_eps": 1e-05,
|
| 10 |
+
"rope_base": 10000.0,
|
| 11 |
+
"rope_scaling": 1.0,
|
| 12 |
+
"max_seq_len": 2048,
|
| 13 |
+
"moe": {
|
| 14 |
+
"even_only": true,
|
| 15 |
+
"num_experts": 8,
|
| 16 |
+
"top_k": 2,
|
| 17 |
+
"capacity_factor": 1.25,
|
| 18 |
+
"eval_capacity_factor": 1.0,
|
| 19 |
+
"noisy_gate_policy": "Jitter",
|
| 20 |
+
"use_residual": false
|
| 21 |
+
},
|
| 22 |
+
"tie_embedding": true,
|
| 23 |
+
"dtype": "bf16"
|
| 24 |
+
}
|