Julien Simon commited on
Commit
fe190a7
1 Parent(s): b63ff12

Add SuperNova on inf2.48xlarge

Browse files
Files changed (2) hide show
  1. results.py +1 -0
  2. results_arcee_supernova.py +8 -0
results.py CHANGED
@@ -27,6 +27,7 @@ instance_type_mappings = {
27
  "c6i.2xlarge": {"cloud": "AWS", "gpu": "-", "gpuRAM": "-"},
28
  "c6i.4xlarge": {"cloud": "AWS", "gpu": "-", "gpuRAM": "-"},
29
  "c7i.4xlarge": {"cloud": "AWS", "gpu": "-", "gpuRAM": "-"},
 
30
  "inf2.*": {"cloud": "AWS", "gpu": "-", "gpuRAM": "-"},
31
  }
32
 
 
27
  "c6i.2xlarge": {"cloud": "AWS", "gpu": "-", "gpuRAM": "-"},
28
  "c6i.4xlarge": {"cloud": "AWS", "gpu": "-", "gpuRAM": "-"},
29
  "c7i.4xlarge": {"cloud": "AWS", "gpu": "-", "gpuRAM": "-"},
30
+ "inf2.48xlarge": {"cloud": "AWS", "gpu": "24xNeuronCore v2", "gpuRAM": "384 GB"},
31
  "inf2.*": {"cloud": "AWS", "gpu": "-", "gpuRAM": "-"},
32
  }
33
 
results_arcee_supernova.py CHANGED
@@ -12,6 +12,14 @@ results_arcee_supernova = {
12
  "tokensPerSecond": "33",
13
  "notes": "",
14
  },
 
 
 
 
 
 
 
 
15
  {
16
  "instanceType": "p4d.24xlarge",
17
  "quantization": "none",
 
12
  "tokensPerSecond": "33",
13
  "notes": "",
14
  },
15
+ {
16
+ "instanceType": "inf2.48xlarge",
17
+ "quantization": "none",
18
+ "container": "transformers-neuronx",
19
+ "status": "OK",
20
+ "tokensPerSecond": "28.4",
21
+ "notes": "bs=4,seqlen=4096",
22
+ },
23
  {
24
  "instanceType": "p4d.24xlarge",
25
  "quantization": "none",