dasLOL's picture
fix_all log
e9e7595 verified
Raw History Blame Contribute Delete
2.11 kB
[12:21:50] phase 1: classify 100% dual-format
[12:46:43] cd2f raw: val 0.9997 test 0.9998 | chat: val 0.9997 test 1.0000
[13:02:42] ca4f raw: val 0.9987 test 0.9978 | chat: val 0.9981
[13:02:54] phase 2: ONNX reduce_range builds
[13:04:13] cd2f onnx_rr: === onnx_cd2f/final_rr on cl_validation_all.jsonl first 200: ACCURACY 1.0000 (4s)
[13:04:34] cd2f calib: {"artifact": "onnx_cd2f/final_rr", "ttft_p50": 73, "ttft_p95": 74, "tok_s": 92.67, "cold_ms": 2156, "rss_MiB": 539, "size_MiB": 339, "requests": 200, "failures"
[13:04:47] staged onnx_cd2f_rr
[13:06:13] ca4f onnx_rr: === onnx_ca4f/final_rr on cl_validation_all.jsonl first 200: ACCURACY 1.0000 (8s)
[13:06:49] ca4f calib: {"artifact": "onnx_ca4f/final_rr", "ttft_p50": 137, "ttft_p95": 138, "tok_s": 70.14, "cold_ms": 2354, "rss_MiB": 580, "size_MiB": 369, "requests": 200, "failure
[13:07:01] staged onnx_ca4f_rr
[13:07:02] phase 3: GGUF q8_0 of cd2f
[13:07:09] gguf converted: 204686592 bytes
[13:07:19] === gguf_cd2f on cl_validation_all.jsonl first 100: ACCURACY 1.0000 | task ttft p50 74 p95 80 ms probe 0: ok=True ttft=0.0 ms total=148 ms tokens=0 output='' probe 1: ok=True ttft=0.0 ms total=166 ms tokens=0 output='' probe 2: ok=True ttft=0.0 ms total=168 ms tokens=0 output=''
[13:07:49] gguf calib: {"artifact": "gguf_cd2f", "ttft_p50": 125, "ttft_p95": 126, "tok_s": 51.69, "cold_ms": 538, "rss_MiB": 430, "size_MiB": 195, "requests": 200, "failures": 0, "wall_s": 29.7}
[13:07:59] staged gguf_cd2f
[13:08:00] phase 4: extract s2/s4 (100%) rebuilt with reduce_range
[13:09:40] s2 onnx_rr: MICRO-F1 1.0000 malformed 0/200
[13:10:08] s2 calib: {"artifact": "onnx_s2/final_rr", "ttft_p50": 74, "ttft_p95": 74, "tok_s": 90.14, "cold_ms": 2198, "rss_MiB": 539, "size_MiB": 339, "requests": 200, "failures":
[13:10:19] staged onnx_s2_rr
[13:12:21] s4 onnx_rr: MICRO-F1 1.0000 malformed 0/200
[13:13:05] s4 calib: {"artifact": "onnx_s4/final_rr", "ttft_p50": 137, "ttft_p95": 137, "tok_s": 72.67, "cold_ms": 2353, "rss_MiB": 580, "size_MiB": 369, "requests": 200, "failures"
[13:13:16] staged onnx_s4_rr
[13:13:16] FIX_ALL_DONE