Buckets:
| { | |
| "combo_slug": "llama8b_l4x1_cb_kernels", | |
| "model": "meta-llama/Llama-3.1-8B-Instruct", | |
| "hardware": "l4x1", | |
| "config_flags": "--continuous-batching --attn-implementation kernels-community/flash-attn2", | |
| "job_id": "6a8d7887dc4d6ae814e84a40", | |
| "status": "success", | |
| "accuracy": 0.840, | |
| "stderr": 0.026, | |
| "total_time_seconds": 438, | |
| "input_tokens": 410414, | |
| "output_tokens": 46083, | |
| "tokens_per_sec": 105.22, | |
| "notes": "Continuous batching + flash-attn2 kernel from kernels-community. ~7.5x faster than the no-CB baseline on the same hardware." | |
| } | |
Xet Storage Details
- Size:
- 560 Bytes
- Xet hash:
- 3f77bd2083dc91e961143a6045613fbfc4b76ebc610eda2d67842b268bafd321
·
Xet efficiently stores files, intelligently splitting them into unique chunks and accelerating uploads and downloads. More info.