Update inference-cache-config/trn1/llama4.json
Browse files
inference-cache-config/trn1/llama4.json
CHANGED
|
@@ -6,7 +6,6 @@
|
|
| 6 |
"batch_size": 1,
|
| 7 |
"sequence_length": 4096,
|
| 8 |
"tensor_parallel_size": 16,
|
| 9 |
-
"auto_cast_type": "bf16"
|
| 10 |
},
|
| 11 |
{
|
| 12 |
"task": "text-generation",
|
|
@@ -14,7 +13,6 @@
|
|
| 14 |
"batch_size": 4,
|
| 15 |
"sequence_length": 4096,
|
| 16 |
"tensor_parallel_size": 16,
|
| 17 |
-
"auto_cast_type": "bf16"
|
| 18 |
},
|
| 19 |
{
|
| 20 |
"task": "text-generation",
|
|
@@ -22,7 +20,6 @@
|
|
| 22 |
"batch_size": 1,
|
| 23 |
"sequence_length": 4096,
|
| 24 |
"tensor_parallel_size": 32,
|
| 25 |
-
"auto_cast_type": "bf16"
|
| 26 |
},
|
| 27 |
{
|
| 28 |
"task": "text-generation",
|
|
@@ -30,7 +27,6 @@
|
|
| 30 |
"batch_size": 4,
|
| 31 |
"sequence_length": 4096,
|
| 32 |
"tensor_parallel_size": 32,
|
| 33 |
-
"auto_cast_type": "bf16"
|
| 34 |
},
|
| 35 |
{
|
| 36 |
"task": "text-generation",
|
|
@@ -38,7 +34,6 @@
|
|
| 38 |
"batch_size": 8,
|
| 39 |
"sequence_length": 4096,
|
| 40 |
"tensor_parallel_size": 32,
|
| 41 |
-
"auto_cast_type": "bf16"
|
| 42 |
}
|
| 43 |
]
|
| 44 |
}
|
|
|
|
| 6 |
"batch_size": 1,
|
| 7 |
"sequence_length": 4096,
|
| 8 |
"tensor_parallel_size": 16,
|
|
|
|
| 9 |
},
|
| 10 |
{
|
| 11 |
"task": "text-generation",
|
|
|
|
| 13 |
"batch_size": 4,
|
| 14 |
"sequence_length": 4096,
|
| 15 |
"tensor_parallel_size": 16,
|
|
|
|
| 16 |
},
|
| 17 |
{
|
| 18 |
"task": "text-generation",
|
|
|
|
| 20 |
"batch_size": 1,
|
| 21 |
"sequence_length": 4096,
|
| 22 |
"tensor_parallel_size": 32,
|
|
|
|
| 23 |
},
|
| 24 |
{
|
| 25 |
"task": "text-generation",
|
|
|
|
| 27 |
"batch_size": 4,
|
| 28 |
"sequence_length": 4096,
|
| 29 |
"tensor_parallel_size": 32,
|
|
|
|
| 30 |
},
|
| 31 |
{
|
| 32 |
"task": "text-generation",
|
|
|
|
| 34 |
"batch_size": 8,
|
| 35 |
"sequence_length": 4096,
|
| 36 |
"tensor_parallel_size": 32,
|
|
|
|
| 37 |
}
|
| 38 |
]
|
| 39 |
}
|