Update README.md
#1
by
helloxm
- opened
README.md
CHANGED
|
@@ -20,6 +20,8 @@ Please follow the [instructions](https://github.com/google-ai-edge/mediapipe-sam
|
|
| 20 |
|
| 21 |
## Benchmarking results
|
| 22 |
|
|
|
|
|
|
|
| 23 |
| Model | Params | | GGML tk/s (CPU, 4 threads) | GGML tk/s (CPU, 8 threads) | LiteRT tk/s (XNNPACK, 4 threads) | LiteRT tk/s (XNNPACK, 8 threads) |
|
| 24 |
| ---------------------------------------------- | ------ | ------------------ | -------------------------- | -------------------------- | -------------------------------- | -------------------------------- |
|
| 25 |
| DeepSeek-R1-Distill-Qwen-1.5B (Int8 quantized) | 1.78 B | Prefill 512 tokens | 64.66 | 87.18 | 260.95 | 299.15 |
|
|
|
|
| 20 |
|
| 21 |
## Benchmarking results
|
| 22 |
|
| 23 |
+
Note that all benchmark stats are from a Samsung S24 Ultra.
|
| 24 |
+
|
| 25 |
| Model | Params | | GGML tk/s (CPU, 4 threads) | GGML tk/s (CPU, 8 threads) | LiteRT tk/s (XNNPACK, 4 threads) | LiteRT tk/s (XNNPACK, 8 threads) |
|
| 26 |
| ---------------------------------------------- | ------ | ------------------ | -------------------------- | -------------------------- | -------------------------------- | -------------------------------- |
|
| 27 |
| DeepSeek-R1-Distill-Qwen-1.5B (Int8 quantized) | 1.78 B | Prefill 512 tokens | 64.66 | 87.18 | 260.95 | 299.15 |
|