Update inference-cache-config/qwen-moe.json
Browse files
inference-cache-config/qwen-moe.json
CHANGED
|
@@ -17,6 +17,12 @@
|
|
| 17 |
"sequence_length": 4096,
|
| 18 |
"num_cores": 16,
|
| 19 |
"auto_cast_type": "bf16"
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 20 |
}
|
| 21 |
-
|
| 22 |
}
|
|
|
|
| 17 |
"sequence_length": 4096,
|
| 18 |
"num_cores": 16,
|
| 19 |
"auto_cast_type": "bf16"
|
| 20 |
+
},
|
| 21 |
+
{
|
| 22 |
+
"batch_size": 8,
|
| 23 |
+
"sequence_length": 4096,
|
| 24 |
+
"num_cores": 32,
|
| 25 |
+
"auto_cast_type": "bf16"
|
| 26 |
}
|
| 27 |
+
]
|
| 28 |
}
|