ByteMLPerf issues | Ecosyste.ms: OpenCollective

Add TPU backend for general_perf and llm_perf

github.com/bytedance/ByteMLPerf - Yi-sir opened this pull request 4 days ago

general_perf厂商接入的环境问题

github.com/bytedance/ByteMLPerf - Yi-sir opened this issue 5 days ago

Gemm a8w8 bench

github.com/bytedance/ByteMLPerf - shengnxu opened this pull request 11 days ago

llm_perf 厂商接入时torch的版本问题

github.com/bytedance/ByteMLPerf - yizhou-xu opened this issue 13 days ago

[CPU]activate_this.py missing

github.com/bytedance/ByteMLPerf - zhengy001 opened this issue 19 days ago

AMD GPU

github.com/bytedance/ByteMLPerf - trinashih opened this issue 23 days ago

【llm_perf】 GPU backend chatglm2-6b，hardcode max_seq_len=4096 when init_kvcache

github.com/bytedance/ByteMLPerf - wufann opened this issue 25 days ago

【llm_perf issue】using byte_infer_perf/llm_perf/launch.py to test chatglm, but meet multi-process competing

github.com/bytedance/ByteMLPerf - danielhua23 opened this issue 30 days ago

[llm_perf] add tensorrt-llm benchmark script.

github.com/bytedance/ByteMLPerf - suisiyuan opened this pull request about 1 month ago

[micro_perf] add barrier to avoid hang.

github.com/bytedance/ByteMLPerf - suisiyuan opened this pull request about 1 month ago

[micro_perf] optimize code structure.

github.com/bytedance/ByteMLPerf - suisiyuan opened this pull request about 1 month ago

[micro_perf] fix log and sqrt in gpu backend.

github.com/bytedance/ByteMLPerf - suisiyuan opened this pull request about 1 month ago

basic ROCM backend, with fused moe

github.com/bytedance/ByteMLPerf - valarLip opened this pull request about 1 month ago

Rocm backend tmp

github.com/bytedance/ByteMLPerf - valarLip opened this pull request about 2 months ago

add single_query and bench_model scripts to debug and perf llm models.

github.com/bytedance/ByteMLPerf - suisiyuan opened this pull request about 2 months ago

llm perf with vllm server

github.com/bytedance/ByteMLPerf - At1a8 opened this issue 2 months ago

[micro_perf] use less memory to avoid using L2 Cache; fix bugs

github.com/bytedance/ByteMLPerf - suisiyuan opened this pull request 2 months ago

[micro perf]: add log and sqrt op

github.com/bytedance/ByteMLPerf - YJessicaGao opened this pull request 2 months ago

[micro_perf] bug_fix, mkdir reports in launch.py.

github.com/bytedance/ByteMLPerf - suisiyuan opened this pull request 2 months ago

[micro_perf] add run log; add process management; fix ops

github.com/bytedance/ByteMLPerf - suisiyuan opened this pull request 2 months ago

provide reports convert script.

github.com/bytedance/ByteMLPerf - suisiyuan opened this pull request 2 months ago

Add Version and Execution Date info to reports

github.com/bytedance/ByteMLPerf - YJessicaGao opened this pull request 3 months ago

add barrier for ccl ops in micro_perf.

github.com/bytedance/ByteMLPerf - suisiyuan opened this pull request 3 months ago

support AMD backend for micro_perf.

github.com/bytedance/ByteMLPerf - suisiyuan opened this pull request 3 months ago

accelerate micro_perf test

github.com/bytedance/ByteMLPerf - suisiyuan opened this pull request 3 months ago

support multi-card perf for micro_perf.

github.com/bytedance/ByteMLPerf - suisiyuan opened this pull request 3 months ago

[llm_perf] fix wrong return value type of CoreInferencer.Task.get_result

github.com/bytedance/ByteMLPerf - qinhj opened this pull request 3 months ago

[micro_perf] fix parsing logic of group gemm

github.com/bytedance/ByteMLPerf - kevinsouthByteDance opened this pull request 3 months ago

Fix gemm: fix workloads of batch/group gemm; add task_dir in perf_engine; multiply qps with 1000

github.com/bytedance/ByteMLPerf - kevinsouthByteDance opened this pull request 3 months ago

enable NPUGroupGemmOp.

github.com/bytedance/ByteMLPerf - wei1tang opened this pull request 3 months ago

fix micro perf launch logic, fix scatter and gather index tensor.

github.com/bytedance/ByteMLPerf - suisiyuan opened this pull request 3 months ago

fix falcon; add scatter, gather to micro_perf

github.com/bytedance/ByteMLPerf - suisiyuan opened this pull request 4 months ago

add llama3-70b, falcon-180b, mixtral8x22b to llm_perf.

github.com/bytedance/ByteMLPerf - suisiyuan opened this pull request 4 months ago

general_perf: remove gpt2, llama2 7b, chatglm2 6b from sota model list

github.com/bytedance/ByteMLPerf - YJessicaGao opened this pull request 4 months ago

microperf: add gemm NK 4k 8k 12k; fix groupgemm get shape prob

github.com/bytedance/ByteMLPerf - YJessicaGao opened this pull request 4 months ago

fix vendor_zoo nvidia cards.

github.com/bytedance/ByteMLPerf - suisiyuan opened this pull request 4 months ago

Musa dev fp16

github.com/bytedance/ByteMLPerf - gliangMT opened this pull request 4 months ago

Musa dev fp16

github.com/bytedance/ByteMLPerf - gonggqing opened this pull request 5 months ago

[llm_perf] fix llm_perf test_accuracy

github.com/bytedance/ByteMLPerf - suisiyuan opened this pull request 5 months ago

[llm_perf] add tp, kvcache and schedule support

github.com/bytedance/ByteMLPerf - suisiyuan opened this pull request 5 months ago

[llm_perf] optimzie llm_perf code structure

github.com/bytedance/ByteMLPerf - suisiyuan opened this pull request 5 months ago

【Issue Help】 run bs=8, input_len=2048 perf error

github.com/bytedance/ByteMLPerf - DeepTecher opened this issue 5 months ago

【Issue Help】 chatglm2-6b has some cases dismatch with golden

github.com/bytedance/ByteMLPerf - DeepTecher opened this issue 5 months ago

【Bug】Bug Need fix -- dismatch output dimension with gloden

github.com/bytedance/ByteMLPerf - DeepTecher opened this issue 5 months ago

add MTGPU backends for operators

github.com/bytedance/ByteMLPerf - qiyulei-mt opened this pull request 5 months ago

add musa adaptation

github.com/bytedance/ByteMLPerf - zehan-mt opened this pull request 6 months ago

显存溢出

github.com/bytedance/ByteMLPerf - incomingflyingbrick opened this issue 6 months ago

python 3.10.12无法启动

github.com/bytedance/ByteMLPerf - incomingflyingbrick opened this issue 6 months ago

有没有可能支持llama3推理测试

github.com/bytedance/ByteMLPerf - incomingflyingbrick opened this issue 6 months ago

[micro_perf] add ops, add int8, fix dist bug.

github.com/bytedance/ByteMLPerf - suisiyuan opened this pull request 6 months ago

chore(micro): Update workloads; Add error message; Support pass task_dir

github.com/bytedance/ByteMLPerf - YJessicaGao opened this pull request 6 months ago

[llm_perf] fix code mistake.

github.com/bytedance/ByteMLPerf - suisiyuan opened this pull request 7 months ago

llm_perf not running

github.com/bytedance/ByteMLPerf - ronysun opened this issue 7 months ago

【ILUVATAR】add general model infer case.

github.com/bytedance/ByteMLPerf - yudefu opened this pull request 7 months ago

[micro_perf] fix host2device process_inputs bug; add tensor shape to micro_perf report.

github.com/bytedance/ByteMLPerf - suisiyuan opened this pull request 7 months ago

[byte_infer_perf] simplify code structure of llm_perf, using separate launch.py as entry point for general_perf and llm_perf.

github.com/bytedance/ByteMLPerf - suisiyuan opened this pull request 7 months ago

fix(general_perf): fix deberta inputs mismatch prob

github.com/bytedance/ByteMLPerf - YJessicaGao opened this pull request 7 months ago

[micro perf] add broadcast op

github.com/bytedance/ByteMLPerf - suisiyuan opened this pull request 7 months ago

Revert "add broadcast op to micro_perf. (#60)"

github.com/bytedance/ByteMLPerf - suisiyuan opened this pull request 7 months ago

[micro perf] add broadcast op.

github.com/bytedance/ByteMLPerf - suisiyuan opened this pull request 7 months ago

【Iluvatar】add iluvatar operator && llm && bert/resnet50/widedeep/yolov5 ixrt infer backend.

github.com/bytedance/ByteMLPerf - forestlee95 opened this pull request 7 months ago

Modify torchrun parameter

github.com/bytedance/ByteMLPerf - davidshtian opened this pull request 8 months ago

(docs) refine micro perf readme