Upload dataset
Browse files
1xA100_EXPERTS_3_QUANT_NAIVE/train-00000-of-00001.parquet
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:f47faeabfbb6d4d7f70a423cf30dcf08d8148fed7d1d51494c46ef85338da2d2
|
3 |
+
size 56291
|
README.md
CHANGED
@@ -524,6 +524,268 @@ dataset_info:
|
|
524 |
num_examples: 7
|
525 |
download_size: 55709
|
526 |
dataset_size: 8920
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
527 |
- config_name: 1xA100_EXPERTS_3_UNQUANT
|
528 |
features:
|
529 |
- name: server_args
|
@@ -1581,6 +1843,10 @@ configs:
|
|
1581 |
data_files:
|
1582 |
- split: train
|
1583 |
path: 1xA100_EXPERTS_2_UNQUANT/train-*
|
|
|
|
|
|
|
|
|
1584 |
- config_name: 1xA100_EXPERTS_3_UNQUANT
|
1585 |
data_files:
|
1586 |
- split: train
|
|
|
524 |
num_examples: 7
|
525 |
download_size: 55709
|
526 |
dataset_size: 8920
|
527 |
+
- config_name: 1xA100_EXPERTS_3_QUANT_NAIVE
|
528 |
+
features:
|
529 |
+
- name: server_args
|
530 |
+
struct:
|
531 |
+
- name: allow_toppings_registration
|
532 |
+
dtype: bool
|
533 |
+
- name: api_key
|
534 |
+
dtype: 'null'
|
535 |
+
- name: attention_backend
|
536 |
+
dtype: string
|
537 |
+
- name: base_min_new_token_ratio
|
538 |
+
dtype: float64
|
539 |
+
- name: chat_template
|
540 |
+
dtype: 'null'
|
541 |
+
- name: chunked_prefill_size
|
542 |
+
dtype: int64
|
543 |
+
- name: constrained_json_whitespace_pattern
|
544 |
+
dtype: 'null'
|
545 |
+
- name: context_length
|
546 |
+
dtype: int64
|
547 |
+
- name: controller_port
|
548 |
+
dtype: int64
|
549 |
+
- name: debug
|
550 |
+
dtype: bool
|
551 |
+
- name: decode_log_interval
|
552 |
+
dtype: int64
|
553 |
+
- name: detokenizer_ipc_name
|
554 |
+
dtype: string
|
555 |
+
- name: detokenizer_port
|
556 |
+
dtype: int64
|
557 |
+
- name: device
|
558 |
+
dtype: string
|
559 |
+
- name: disable_cuda_graph
|
560 |
+
dtype: bool
|
561 |
+
- name: disable_cuda_graph_padding
|
562 |
+
dtype: bool
|
563 |
+
- name: disable_custom_all_reduce
|
564 |
+
dtype: bool
|
565 |
+
- name: disable_disk_cache
|
566 |
+
dtype: bool
|
567 |
+
- name: disable_flashinfer
|
568 |
+
dtype: bool
|
569 |
+
- name: disable_flashinfer_sampling
|
570 |
+
dtype: bool
|
571 |
+
- name: disable_mla
|
572 |
+
dtype: bool
|
573 |
+
- name: disable_nan_detection
|
574 |
+
dtype: bool
|
575 |
+
- name: disable_penalizer
|
576 |
+
dtype: bool
|
577 |
+
- name: disable_radix_cache
|
578 |
+
dtype: bool
|
579 |
+
- name: disable_regex_jump_forward
|
580 |
+
dtype: bool
|
581 |
+
- name: dist_init_addr
|
582 |
+
dtype: 'null'
|
583 |
+
- name: dp_size
|
584 |
+
dtype: int64
|
585 |
+
- name: dtype
|
586 |
+
dtype: string
|
587 |
+
- name: enable_double_sparsity
|
588 |
+
dtype: bool
|
589 |
+
- name: enable_dp_attention
|
590 |
+
dtype: bool
|
591 |
+
- name: enable_mixed_chunk
|
592 |
+
dtype: bool
|
593 |
+
- name: enable_overlap_schedule
|
594 |
+
dtype: bool
|
595 |
+
- name: enable_p2p_check
|
596 |
+
dtype: bool
|
597 |
+
- name: enable_stats_logging
|
598 |
+
dtype: bool
|
599 |
+
- name: enable_system_controller
|
600 |
+
dtype: bool
|
601 |
+
- name: enable_toppings
|
602 |
+
dtype: bool
|
603 |
+
- name: enable_torch_compile
|
604 |
+
dtype: bool
|
605 |
+
- name: flashinfer_workspace_size
|
606 |
+
dtype: int64
|
607 |
+
- name: host
|
608 |
+
dtype: string
|
609 |
+
- name: init_new_token_ratio
|
610 |
+
dtype: float64
|
611 |
+
- name: init_number_of_deltas
|
612 |
+
dtype: int64
|
613 |
+
- name: init_number_of_loras
|
614 |
+
dtype: int64
|
615 |
+
- name: init_toppings
|
616 |
+
dtype: 'null'
|
617 |
+
- name: is_embedding
|
618 |
+
dtype: bool
|
619 |
+
- name: kv_cache_dtype
|
620 |
+
dtype: string
|
621 |
+
- name: load_balance_method
|
622 |
+
dtype: string
|
623 |
+
- name: load_format
|
624 |
+
dtype: string
|
625 |
+
- name: log_requests
|
626 |
+
dtype: bool
|
627 |
+
- name: lora_paths
|
628 |
+
dtype: 'null'
|
629 |
+
- name: max_lora_ranks
|
630 |
+
dtype: int64
|
631 |
+
- name: max_prefill_tokens
|
632 |
+
dtype: int64
|
633 |
+
- name: max_running_requests
|
634 |
+
dtype: 'null'
|
635 |
+
- name: max_toppings_per_batch
|
636 |
+
dtype: int64
|
637 |
+
- name: max_torch_compile_bs
|
638 |
+
dtype: int64
|
639 |
+
- name: max_total_tokens
|
640 |
+
dtype: 'null'
|
641 |
+
- name: mem_fraction_static
|
642 |
+
dtype: float64
|
643 |
+
- name: model_path
|
644 |
+
dtype: string
|
645 |
+
- name: nccl_ports
|
646 |
+
sequence: string
|
647 |
+
- name: new_token_ratio_decay
|
648 |
+
dtype: float64
|
649 |
+
- name: nnodes
|
650 |
+
dtype: int64
|
651 |
+
- name: node_rank
|
652 |
+
dtype: int64
|
653 |
+
- name: num_continue_decode_steps
|
654 |
+
dtype: int64
|
655 |
+
- name: num_continuous_decode_steps
|
656 |
+
dtype: int64
|
657 |
+
- name: port
|
658 |
+
dtype: int64
|
659 |
+
- name: quantization
|
660 |
+
dtype: 'null'
|
661 |
+
- name: random_seed
|
662 |
+
dtype: int64
|
663 |
+
- name: retract_decode_steps
|
664 |
+
dtype: int64
|
665 |
+
- name: sampling_backend
|
666 |
+
dtype: string
|
667 |
+
- name: schedule_conservativeness
|
668 |
+
dtype: float64
|
669 |
+
- name: schedule_policy
|
670 |
+
dtype: string
|
671 |
+
- name: scheduler_input_ipc_name
|
672 |
+
dtype: string
|
673 |
+
- name: scheduler_port
|
674 |
+
dtype: int64
|
675 |
+
- name: served_model_name
|
676 |
+
dtype: string
|
677 |
+
- name: server_id
|
678 |
+
dtype: string
|
679 |
+
- name: show_time_cost
|
680 |
+
dtype: bool
|
681 |
+
- name: skip_tokenizer_init
|
682 |
+
dtype: bool
|
683 |
+
- name: stream_interval
|
684 |
+
dtype: int64
|
685 |
+
- name: tokenizer_ipc_name
|
686 |
+
dtype: string
|
687 |
+
- name: tokenizer_mode
|
688 |
+
dtype: string
|
689 |
+
- name: tokenizer_path
|
690 |
+
dtype: string
|
691 |
+
- name: tokenizer_port
|
692 |
+
dtype: int64
|
693 |
+
- name: torchao_config
|
694 |
+
dtype: string
|
695 |
+
- name: tp_size
|
696 |
+
dtype: int64
|
697 |
+
- name: triton_attention_reduce_in_fp32
|
698 |
+
dtype: bool
|
699 |
+
- name: trust_remote_code
|
700 |
+
dtype: bool
|
701 |
+
- name: use_heterogeneous_pool
|
702 |
+
dtype: bool
|
703 |
+
- name: watchdog_timeout
|
704 |
+
dtype: float64
|
705 |
+
- name: client_args
|
706 |
+
struct:
|
707 |
+
- name: dataset
|
708 |
+
dtype: string
|
709 |
+
- name: endpoint
|
710 |
+
dtype: string
|
711 |
+
- name: goodput
|
712 |
+
dtype: 'null'
|
713 |
+
- name: metric_percentiles
|
714 |
+
dtype: string
|
715 |
+
- name: model
|
716 |
+
dtype: string
|
717 |
+
- name: num_prompts
|
718 |
+
dtype: int64
|
719 |
+
- name: output
|
720 |
+
dtype: string
|
721 |
+
- name: percentile_metrics
|
722 |
+
dtype: string
|
723 |
+
- name: request_rate
|
724 |
+
dtype: float64
|
725 |
+
- name: tokenizer
|
726 |
+
dtype: string
|
727 |
+
- name: wait_until_ready
|
728 |
+
dtype: bool
|
729 |
+
- name: metrics
|
730 |
+
struct:
|
731 |
+
- name: completed
|
732 |
+
dtype: int64
|
733 |
+
- name: mean_e2el_ms
|
734 |
+
dtype: float64
|
735 |
+
- name: mean_itl_ms
|
736 |
+
dtype: float64
|
737 |
+
- name: mean_tpot_ms
|
738 |
+
dtype: float64
|
739 |
+
- name: mean_ttft_ms
|
740 |
+
dtype: float64
|
741 |
+
- name: median_e2el_ms
|
742 |
+
dtype: float64
|
743 |
+
- name: median_itl_ms
|
744 |
+
dtype: float64
|
745 |
+
- name: median_tpot_ms
|
746 |
+
dtype: float64
|
747 |
+
- name: median_ttft_ms
|
748 |
+
dtype: float64
|
749 |
+
- name: output_throughput
|
750 |
+
dtype: float64
|
751 |
+
- name: percentiles_e2el_ms
|
752 |
+
sequence:
|
753 |
+
sequence: float64
|
754 |
+
- name: percentiles_itl_ms
|
755 |
+
sequence:
|
756 |
+
sequence: float64
|
757 |
+
- name: percentiles_tpot_ms
|
758 |
+
sequence:
|
759 |
+
sequence: float64
|
760 |
+
- name: percentiles_ttft_ms
|
761 |
+
sequence:
|
762 |
+
sequence: float64
|
763 |
+
- name: request_goodput
|
764 |
+
dtype: float64
|
765 |
+
- name: request_throughput
|
766 |
+
dtype: float64
|
767 |
+
- name: std_e2el_ms
|
768 |
+
dtype: float64
|
769 |
+
- name: std_itl_ms
|
770 |
+
dtype: float64
|
771 |
+
- name: std_tpot_ms
|
772 |
+
dtype: float64
|
773 |
+
- name: std_ttft_ms
|
774 |
+
dtype: float64
|
775 |
+
- name: total_input
|
776 |
+
dtype: int64
|
777 |
+
- name: total_output
|
778 |
+
dtype: int64
|
779 |
+
- name: total_token_throughput
|
780 |
+
dtype: float64
|
781 |
+
- name: device
|
782 |
+
dtype: string
|
783 |
+
splits:
|
784 |
+
- name: train
|
785 |
+
num_bytes: 9732
|
786 |
+
num_examples: 7
|
787 |
+
download_size: 56291
|
788 |
+
dataset_size: 9732
|
789 |
- config_name: 1xA100_EXPERTS_3_UNQUANT
|
790 |
features:
|
791 |
- name: server_args
|
|
|
1843 |
data_files:
|
1844 |
- split: train
|
1845 |
path: 1xA100_EXPERTS_2_UNQUANT/train-*
|
1846 |
+
- config_name: 1xA100_EXPERTS_3_QUANT_NAIVE
|
1847 |
+
data_files:
|
1848 |
+
- split: train
|
1849 |
+
path: 1xA100_EXPERTS_3_QUANT_NAIVE/train-*
|
1850 |
- config_name: 1xA100_EXPERTS_3_UNQUANT
|
1851 |
data_files:
|
1852 |
- split: train
|