Share your thoughts, 1 month free Claude Pro on us
See more
Home
/
Benchmarks
Decoding Throughput on Alpaca
Loading...
88.8
AR Throughput (tok/s)
Autoregressive
53.336
62.543
71.75
80.957
Mar 3, 2026
AR Throughput (tok/s)
SD Throughput (tok/s)
SSD Throughput (tok/s)
SSD/SD Speedup
SSD/AR Speedup
Updated 1mo ago
Evaluation Results
Method
Method
Links
AR Throughput (tok/s)
SD Throughput (tok/s)
SSD Throughput (tok/s)
SSD/SD Speedup
SSD/AR Speedup
Autoregressive
Model=Qwen-3 32B/0.6B
2026.03
88.8
-
-
-
-
Autoregressive
Model=Llama-3.1-Instru...
2026.03
54.7
-
-
-
-
Speculative Decoding
Model=Llama-3.1-Instru...
2026.03
-
145
-
-
-
SSD
Model=Llama-3.1-Instru...
2026.03
-
-
224
1.55
4.1
Speculative Decoding
Model=Qwen-3 32B/0.6B
2026.03
-
127
-
-
-
SSD
Model=Qwen-3 32B/0.6B
2026.03
-
-
185
1.47
2.08
Feedback
Search any
task
Search any
task