Hugging Face's logo

pytorch
/
Qwen3-32B-FP8

Text Generation
Transformers
PyTorch
multilingual
qwen3
torchao
code
math
chat
conversational
text-generation-inference
Model card Files
xet
Community
SocialLocalMobile commited on
Commit
3b10b05
·
verified ·
1 Parent(s): edf5c51

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +3 -5
README.md CHANGED
@@ -129,9 +129,7 @@ TODO
129
 
130
  # Memory Usage
131
 
132
- Tested on H100
133
-
134
- | Memory | | |
135
  |----------------------------------|----------------|-------------------------------|
136
  | | Qwen3-32B | Qwen3-32B-float8dq |
137
  | Peak Memory | 65.72 GB | 34.54 GB (-47.44%) |
@@ -198,8 +196,8 @@ print(f"Peak Memory Usage: {mem:.02f} GB")
198
 
199
  # Model Performance
200
 
201
- Tested on H100
202
- | Benchmark | | |
203
  |----------------------------------|----------------|-------------------------------|
204
  | | Qwen3-32B | Qwen3-32B-float8dq |
205
  | latency (batch_size=1) | 9.1s | 5.77s (-36.6%) |
 
129
 
130
  # Memory Usage
131
 
132
+ | Memory (tested on H100) | | |
 
 
133
  |----------------------------------|----------------|-------------------------------|
134
  | | Qwen3-32B | Qwen3-32B-float8dq |
135
  | Peak Memory | 65.72 GB | 34.54 GB (-47.44%) |
 
196
 
197
  # Model Performance
198
 
199
+
200
+ | Benchmark (Tested on H100) | | |
201
  |----------------------------------|----------------|-------------------------------|
202
  | | Qwen3-32B | Qwen3-32B-float8dq |
203
  | latency (batch_size=1) | 9.1s | 5.77s (-36.6%) |