Upload config.json
Browse files- config.json +12 -6
config.json
CHANGED
|
@@ -6,15 +6,20 @@
|
|
| 6 |
],
|
| 7 |
"bos_token_id": 50256,
|
| 8 |
"eos_token_id": 50256,
|
| 9 |
-
"
|
| 10 |
-
"
|
| 11 |
-
"
|
| 12 |
-
"
|
| 13 |
-
"
|
| 14 |
"n_ctx": 1024,
|
| 15 |
"n_positions": 2048,
|
| 16 |
"n_special": 0,
|
| 17 |
"num_moe_experts": 16,
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 18 |
"pad_token_id": 50256,
|
| 19 |
"predict_special_tokens": true,
|
| 20 |
"reward_loss_weight": 0.1,
|
|
@@ -29,5 +34,6 @@
|
|
| 29 |
}
|
| 30 |
},
|
| 31 |
"torch_dtype": "float32",
|
| 32 |
-
"vocab_size": 50308
|
|
|
|
| 33 |
}
|
|
|
|
| 6 |
],
|
| 7 |
"bos_token_id": 50256,
|
| 8 |
"eos_token_id": 50256,
|
| 9 |
+
"hidden_size": 1024,
|
| 10 |
+
"n_embd": 1024,
|
| 11 |
+
"n_head": 16,
|
| 12 |
+
"n_layer": 24,
|
| 13 |
+
"n_inner": 4096,
|
| 14 |
"n_ctx": 1024,
|
| 15 |
"n_positions": 2048,
|
| 16 |
"n_special": 0,
|
| 17 |
"num_moe_experts": 16,
|
| 18 |
+
"moe_load_balancing_weight": 0.01,
|
| 19 |
+
"moe_router_z_loss_weight": 0.001,
|
| 20 |
+
"moe_top_k": 4,
|
| 21 |
+
"max_neurons": 8192,
|
| 22 |
+
"max_position_embeddings": 1024,
|
| 23 |
"pad_token_id": 50256,
|
| 24 |
"predict_special_tokens": true,
|
| 25 |
"reward_loss_weight": 0.1,
|
|
|
|
| 34 |
}
|
| 35 |
},
|
| 36 |
"torch_dtype": "float32",
|
| 37 |
+
"vocab_size": 50308,
|
| 38 |
+
"model_type": "gpt2"
|
| 39 |
}
|