Spaces:
Running
Running
Julien Simon
commited on
Commit
•
534766c
1
Parent(s):
d0c0894
More SuperNova
Browse files
results_arcee_supernova.py
CHANGED
|
@@ -50,7 +50,7 @@ results_arcee_supernova = {
|
|
| 50 |
"container": "LMI 0.29+transformers-neuronx 0.11.351",
|
| 51 |
"status": "???",
|
| 52 |
"tokensPerSecond": "???",
|
| 53 |
-
"notes": "bs=2,seqlen=8192 - SDK 2.19.1",
|
| 54 |
},
|
| 55 |
],
|
| 56 |
},
|
|
@@ -83,7 +83,7 @@ results_arcee_supernova = {
|
|
| 83 |
"container": "LMI 0.29+transformers-neuronx 0.11.351",
|
| 84 |
"status": "OK",
|
| 85 |
"tokensPerSecond": "14",
|
| 86 |
-
"notes": "bs=2,seqlen=8192 - SDK 2.19.1",
|
| 87 |
},
|
| 88 |
],
|
| 89 |
},
|
|
@@ -102,7 +102,7 @@ results_arcee_supernova = {
|
|
| 102 |
"container": "LMI 0.29+transformers-neuronx 0.11.351",
|
| 103 |
"status": "OK",
|
| 104 |
"tokensPerSecond": "24.6",
|
| 105 |
-
"notes": "bs=2,seqlen=8192 - SDK 2.19.1",
|
| 106 |
},
|
| 107 |
],
|
| 108 |
},
|
|
|
|
| 50 |
"container": "LMI 0.29+transformers-neuronx 0.11.351",
|
| 51 |
"status": "???",
|
| 52 |
"tokensPerSecond": "???",
|
| 53 |
+
"notes": "bs=2,seqlen=8192 - SDK 2.19.1 - OPTION_LOAD_IN_8BIT=True",
|
| 54 |
},
|
| 55 |
],
|
| 56 |
},
|
|
|
|
| 83 |
"container": "LMI 0.29+transformers-neuronx 0.11.351",
|
| 84 |
"status": "OK",
|
| 85 |
"tokensPerSecond": "14",
|
| 86 |
+
"notes": "bs=2,seqlen=8192 - SDK 2.19.1 - OPTION_LOAD_IN_8BIT=True",
|
| 87 |
},
|
| 88 |
],
|
| 89 |
},
|
|
|
|
| 102 |
"container": "LMI 0.29+transformers-neuronx 0.11.351",
|
| 103 |
"status": "OK",
|
| 104 |
"tokensPerSecond": "24.6",
|
| 105 |
+
"notes": "bs=2,seqlen=8192 - SDK 2.19.1 - OPTION_LOAD_IN_8BIT=True",
|
| 106 |
},
|
| 107 |
],
|
| 108 |
},
|
results_llama_supernova_lite.py
CHANGED
|
@@ -4,6 +4,13 @@ results_llama_supernova_lite = {
|
|
| 4 |
"name": "Llama-3.1-SuperNova-Lite",
|
| 5 |
"modelType": "Llama 3.1 8B",
|
| 6 |
"configurations": [
|
| 7 |
-
{
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 8 |
],
|
| 9 |
}
|
|
|
|
| 4 |
"name": "Llama-3.1-SuperNova-Lite",
|
| 5 |
"modelType": "Llama 3.1 8B",
|
| 6 |
"configurations": [
|
| 7 |
+
{
|
| 8 |
+
"instanceType": "c7g.8xlarge",
|
| 9 |
+
"quantization": "Q4_0_8_8",
|
| 10 |
+
"container": "llama.cpp 9/18/24",
|
| 11 |
+
"status": "OK",
|
| 12 |
+
"tokensPerSecond": "39.7",
|
| 13 |
+
"notes": "requantized from Q4_K_S",
|
| 14 |
+
},
|
| 15 |
],
|
| 16 |
}
|