Datasets:

masterpieceexternal
/

gpt-oss-20b-lmsys-layer2-expert-traces-600k-10msps

Tags:

License:

Dataset card Files Files and versions

xet

Community

Dataset Viewer

Duplicate

Split (1)

train

The dataset viewer is not available for this split.

Cannot load the dataset split (in streaming mode) to extract the first rows.

Error code:   StreamingRowsError
Exception:    CastError
Message:      Couldn't cast
mode: string
backend: string
dataset: string
split: string
text_key: string
start: int64
limit: int64
shuffle: bool
local_files_only: bool
streaming: bool
shuffle_buffer: int64
prompt_language: string
min_source_prompt_tokens: int64
max_source_prompt_tokens: int64
dedupe_prompts: bool
skip_generic_greetings: bool
exclude_redacted: bool
rows_scanned_after_start: int64
rows_seen_total: int64
reject_counts: struct<empty: int64, language: int64, redacted: int64, too_short: int64, too_long: int64, generic_gr (... 32 chars omitted)
  child 0, empty: int64
  child 1, language: int64
  child 2, redacted: int64
  child 3, too_short: int64
  child 4, too_long: int64
  child 5, generic_greeting: int64
  child 6, duplicate: int64
num_prompts_loaded: int64
loaded_prompt_hashes: list<item: string>
  child 0, item: string
loaded_source_row_indices: list<item: int64>
  child 0, item: int64
loaded_prompt_languages: list<item: string>
  child 0, item: string
loaded_prompt_token_counts: list<item: int64>
  child 0, item: int64
tokens_per_prompt: int64
num_prompts_used: int64
decode_steps_requested: int64
captured_region: string
gpu_end: struct<power_W: double, sm_clock_MHz: int64, mem_clock_MHz: int64, temp_C: int64>
  child 0, power_W: double
  child 1, sm_clock_MHz: int64
  child 2, mem_clock_MHz: int64
  child 3, temp_C: int64
prompt_dataset: string
prompt_meta: struct<mode: string, backend: string, dataset: string, split: string, text_key: string, start: int64 (... 756 chars omitted)
  chil
...
hild 27, decode_steps_requested: int64
rejected_attempt_count: int64
scope: struct<adc_freq_raw: double, adc_decimate: int64, adc_freq_effective: double, adc_samples: int64, ca (... 97 chars omitted)
  child 0, adc_freq_raw: double
  child 1, adc_decimate: int64
  child 2, adc_freq_effective: double
  child 3, adc_samples: int64
  child 4, capture_ms_effective: double
  child 5, stream_mode: bool
  child 6, adc_bits_per_sample: int64
  child 7, max_samples_hw: int64
stopped_cleanly: bool
created_at: string
run_dir: string
timing: struct<model_load_s: double, warmup_s: double, capture_loop_s: double, accepted_per_second_excluding (... 78 chars omitted)
  child 0, model_load_s: double
  child 1, warmup_s: double
  child 2, capture_loop_s: double
  child 3, accepted_per_second_excluding_load_warmup: double
  child 4, prompt_prefill_count: int64
  child 5, prompt_replay_count: int64
accepted_count: int64
captured_layers: list<item: int64>
  child 0, item: int64
validation: struct<max_abs_mean: double, min_std: double, max_std: double, rail_low_threshold: double, rail_high (... 69 chars omitted)
  child 0, max_abs_mean: double
  child 1, min_std: double
  child 2, max_std: double
  child 3, rail_low_threshold: double
  child 4, rail_high_threshold: double
  child 5, max_rail_frac: double
  child 6, min_expert_windows: int64
off_trace_post: list<item: string>
  child 0, item: string
target_layer_idx: int64
model_name: string
off_trace_pre: list<item: string>
  child 0, item: string
to
{'mode': Value('string'), 'created_at': Value('string'), 'run_dir': Value('string'), 'model_name': Value('string'), 'prompt_dataset': Value('string'), 'prompt_meta': {'mode': Value('string'), 'backend': Value('string'), 'dataset': Value('string'), 'split': Value('string'), 'text_key': Value('string'), 'start': Value('int64'), 'limit': Value('int64'), 'shuffle': Value('bool'), 'local_files_only': Value('bool'), 'streaming': Value('bool'), 'shuffle_buffer': Value('int64'), 'prompt_language': Value('string'), 'min_source_prompt_tokens': Value('int64'), 'max_source_prompt_tokens': Value('int64'), 'dedupe_prompts': Value('bool'), 'skip_generic_greetings': Value('bool'), 'exclude_redacted': Value('bool'), 'rows_scanned_after_start': Value('int64'), 'rows_seen_total': Value('int64'), 'reject_counts': {'empty': Value('int64'), 'language': Value('int64'), 'redacted': Value('int64'), 'too_short': Value('int64'), 'too_long': Value('int64'), 'generic_greeting': Value('int64'), 'duplicate': Value('int64')}, 'num_prompts_loaded': Value('int64'), 'loaded_prompt_hashes': List(Value('string')), 'loaded_source_row_indices': List(Value('int64')), 'loaded_prompt_languages': List(Value('string')), 'loaded_prompt_token_counts': List(Value('int64')), 'tokens_per_prompt': Value('int64'), 'num_prompts_used': Value('int64'), 'decode_steps_requested': Value('int64')}, 'target_layer_idx': Value('int64'), 'captured_region': Value('string'), 'captured_layers': List(Value('int64')), 'off_trace_pre': List(Value('string')), 'off_trace_post': List(Value('string')), 'decode_steps_requested': Value('int64'), 'accepted_count': Value('int64'), 'rejected_attempt_count': Value('int64'), 'timing': {'model_load_s': Value('float64'), 'warmup_s': Value('float64'), 'capture_loop_s': Value('float64'), 'accepted_per_second_excluding_load_warmup': Value('float64'), 'prompt_prefill_count': Value('int64'), 'prompt_replay_count': Value('int64')}, 'scope': {'adc_freq_raw': Value('float64'), 'adc_decimate': Value('int64'), 'adc_freq_effective': Value('float64'), 'adc_samples': Value('int64'), 'capture_ms_effective': Value('float64'), 'stream_mode': Value('bool'), 'adc_bits_per_sample': Value('int64'), 'max_samples_hw': Value('int64')}, 'validation': {'max_abs_mean': Value('float64'), 'min_std': Value('float64'), 'max_std': Value('float64'), 'rail_low_threshold': Value('float64'), 'rail_high_threshold': Value('float64'), 'max_rail_frac': Value('float64'), 'min_expert_windows': Value('int64')}, 'gpu_end': {'power_W': Value('float64'), 'sm_clock_MHz': Value('int64'), 'mem_clock_MHz': Value('int64'), 'temp_C': Value('int64')}, 'stopped_cleanly': Value('bool')}
because column names don't match
Traceback:    Traceback (most recent call last):
                File "/src/services/worker/src/worker/utils.py", line 147, in get_rows_or_raise
                  return get_rows(
                      dataset=dataset,
                  ...<4 lines>...
                      column_names=column_names,
                  )
                File "/src/libs/libcommon/src/libcommon/utils.py", line 272, in decorator
                  return func(*args, **kwargs)
                File "/src/services/worker/src/worker/utils.py", line 127, in get_rows
                  rows_plus_one = list(itertools.islice(safe_iter(ds, dataset=dataset), rows_max_number + 1))
                File "/src/services/worker/src/worker/utils.py", line 478, in safe_iter
                  yield from ds.decode(False) if ds.features else ds
                File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 2818, in __iter__
                  for key, example in ex_iterable:
                                      ^^^^^^^^^^^
                File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 2355, in __iter__
                  for key, pa_table in self._iter_arrow():
                                       ~~~~~~~~~~~~~~~~^^
                File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 2380, in _iter_arrow
                  for key, pa_table in self.ex_iterable._iter_arrow():
                                       ~~~~~~~~~~~~~~~~~~~~~~~~~~~~^^
                File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 536, in _iter_arrow
                  for key, pa_table in iterator:
                                       ^^^^^^^^
                File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 419, in _iter_arrow
                  for key, pa_table in self.generate_tables_fn(**gen_kwags):
                                       ~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^
                File "/usr/local/lib/python3.14/site-packages/datasets/packaged_modules/json/json.py", line 343, in _generate_tables
                  self._cast_table(pa_table, json_field_paths=json_field_paths),
                  ~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
                File "/usr/local/lib/python3.14/site-packages/datasets/packaged_modules/json/json.py", line 132, in _cast_table
                  pa_table = table_cast(pa_table, self.info.features.arrow_schema)
                File "/usr/local/lib/python3.14/site-packages/datasets/table.py", line 2369, in table_cast
                  return cast_table_to_schema(table, schema)
                File "/usr/local/lib/python3.14/site-packages/datasets/table.py", line 2297, in cast_table_to_schema
                  raise CastError(
                  ...<3 lines>...
                  )
              datasets.table.CastError: Couldn't cast
              mode: string
              backend: string
              dataset: string
              split: string
              text_key: string
              start: int64
              limit: int64
              shuffle: bool
              local_files_only: bool
              streaming: bool
              shuffle_buffer: int64
              prompt_language: string
              min_source_prompt_tokens: int64
              max_source_prompt_tokens: int64
              dedupe_prompts: bool
              skip_generic_greetings: bool
              exclude_redacted: bool
              rows_scanned_after_start: int64
              rows_seen_total: int64
              reject_counts: struct<empty: int64, language: int64, redacted: int64, too_short: int64, too_long: int64, generic_gr (... 32 chars omitted)
                child 0, empty: int64
                child 1, language: int64
                child 2, redacted: int64
                child 3, too_short: int64
                child 4, too_long: int64
                child 5, generic_greeting: int64
                child 6, duplicate: int64
              num_prompts_loaded: int64
              loaded_prompt_hashes: list<item: string>
                child 0, item: string
              loaded_source_row_indices: list<item: int64>
                child 0, item: int64
              loaded_prompt_languages: list<item: string>
                child 0, item: string
              loaded_prompt_token_counts: list<item: int64>
                child 0, item: int64
              tokens_per_prompt: int64
              num_prompts_used: int64
              decode_steps_requested: int64
              captured_region: string
              gpu_end: struct<power_W: double, sm_clock_MHz: int64, mem_clock_MHz: int64, temp_C: int64>
                child 0, power_W: double
                child 1, sm_clock_MHz: int64
                child 2, mem_clock_MHz: int64
                child 3, temp_C: int64
              prompt_dataset: string
              prompt_meta: struct<mode: string, backend: string, dataset: string, split: string, text_key: string, start: int64 (... 756 chars omitted)
                chil
              ...
              hild 27, decode_steps_requested: int64
              rejected_attempt_count: int64
              scope: struct<adc_freq_raw: double, adc_decimate: int64, adc_freq_effective: double, adc_samples: int64, ca (... 97 chars omitted)
                child 0, adc_freq_raw: double
                child 1, adc_decimate: int64
                child 2, adc_freq_effective: double
                child 3, adc_samples: int64
                child 4, capture_ms_effective: double
                child 5, stream_mode: bool
                child 6, adc_bits_per_sample: int64
                child 7, max_samples_hw: int64
              stopped_cleanly: bool
              created_at: string
              run_dir: string
              timing: struct<model_load_s: double, warmup_s: double, capture_loop_s: double, accepted_per_second_excluding (... 78 chars omitted)
                child 0, model_load_s: double
                child 1, warmup_s: double
                child 2, capture_loop_s: double
                child 3, accepted_per_second_excluding_load_warmup: double
                child 4, prompt_prefill_count: int64
                child 5, prompt_replay_count: int64
              accepted_count: int64
              captured_layers: list<item: int64>
                child 0, item: int64
              validation: struct<max_abs_mean: double, min_std: double, max_std: double, rail_low_threshold: double, rail_high (... 69 chars omitted)
                child 0, max_abs_mean: double
                child 1, min_std: double
                child 2, max_std: double
                child 3, rail_low_threshold: double
                child 4, rail_high_threshold: double
                child 5, max_rail_frac: double
                child 6, min_expert_windows: int64
              off_trace_post: list<item: string>
                child 0, item: string
              target_layer_idx: int64
              model_name: string
              off_trace_pre: list<item: string>
                child 0, item: string
              to
              {'mode': Value('string'), 'created_at': Value('string'), 'run_dir': Value('string'), 'model_name': Value('string'), 'prompt_dataset': Value('string'), 'prompt_meta': {'mode': Value('string'), 'backend': Value('string'), 'dataset': Value('string'), 'split': Value('string'), 'text_key': Value('string'), 'start': Value('int64'), 'limit': Value('int64'), 'shuffle': Value('bool'), 'local_files_only': Value('bool'), 'streaming': Value('bool'), 'shuffle_buffer': Value('int64'), 'prompt_language': Value('string'), 'min_source_prompt_tokens': Value('int64'), 'max_source_prompt_tokens': Value('int64'), 'dedupe_prompts': Value('bool'), 'skip_generic_greetings': Value('bool'), 'exclude_redacted': Value('bool'), 'rows_scanned_after_start': Value('int64'), 'rows_seen_total': Value('int64'), 'reject_counts': {'empty': Value('int64'), 'language': Value('int64'), 'redacted': Value('int64'), 'too_short': Value('int64'), 'too_long': Value('int64'), 'generic_greeting': Value('int64'), 'duplicate': Value('int64')}, 'num_prompts_loaded': Value('int64'), 'loaded_prompt_hashes': List(Value('string')), 'loaded_source_row_indices': List(Value('int64')), 'loaded_prompt_languages': List(Value('string')), 'loaded_prompt_token_counts': List(Value('int64')), 'tokens_per_prompt': Value('int64'), 'num_prompts_used': Value('int64'), 'decode_steps_requested': Value('int64')}, 'target_layer_idx': Value('int64'), 'captured_region': Value('string'), 'captured_layers': List(Value('int64')), 'off_trace_pre': List(Value('string')), 'off_trace_post': List(Value('string')), 'decode_steps_requested': Value('int64'), 'accepted_count': Value('int64'), 'rejected_attempt_count': Value('int64'), 'timing': {'model_load_s': Value('float64'), 'warmup_s': Value('float64'), 'capture_loop_s': Value('float64'), 'accepted_per_second_excluding_load_warmup': Value('float64'), 'prompt_prefill_count': Value('int64'), 'prompt_replay_count': Value('int64')}, 'scope': {'adc_freq_raw': Value('float64'), 'adc_decimate': Value('int64'), 'adc_freq_effective': Value('float64'), 'adc_samples': Value('int64'), 'capture_ms_effective': Value('float64'), 'stream_mode': Value('bool'), 'adc_bits_per_sample': Value('int64'), 'max_samples_hw': Value('int64')}, 'validation': {'max_abs_mean': Value('float64'), 'min_std': Value('float64'), 'max_std': Value('float64'), 'rail_low_threshold': Value('float64'), 'rail_high_threshold': Value('float64'), 'max_rail_frac': Value('float64'), 'min_expert_windows': Value('int64')}, 'gpu_end': {'power_W': Value('float64'), 'sm_clock_MHz': Value('int64'), 'mem_clock_MHz': Value('int64'), 'temp_C': Value('int64')}, 'stopped_cleanly': Value('bool')}
              because column names don't match

Need help to make the dataset viewer work? Make sure to review how to configure the dataset viewer, and open a discussion for direct support.

YAML Metadata Warning:The task_categories "time-series-classification" is not in the official list: text-classification, token-classification, table-question-answering, question-answering, zero-shot-classification, translation, summarization, feature-extraction, text-generation, fill-mask, sentence-similarity, text-to-speech, text-to-audio, automatic-speech-recognition, audio-to-audio, audio-classification, audio-text-to-text, voice-activity-detection, depth-estimation, image-classification, object-detection, image-segmentation, text-to-image, image-to-text, image-to-image, image-to-video, unconditional-image-generation, video-classification, reinforcement-learning, robotics, tabular-classification, tabular-regression, tabular-to-text, table-to-text, multiple-choice, text-ranking, text-retrieval, time-series-forecasting, text-to-video, image-text-to-text, image-text-to-image, image-text-to-video, visual-question-answering, document-question-answering, zero-shot-image-classification, graph-ml, mask-generation, zero-shot-object-detection, text-to-3d, image-to-3d, image-feature-extraction, video-text-to-text, keypoint-detection, visual-document-retrieval, any-to-any, video-to-video, other

GPT-OSS-20B LMSYS Layer-2 Expert Power Traces — 600k tokens at 10 MSPS

This dataset contains real power/current traces captured from an NVIDIA H100 PCIe system while running GPT-OSS-20B decode on LMSYS prompts. The capture target is layer 2 MoE expert execution only; earlier/later transformer work is executed off-trace to avoid wasting acquisition time.

Capture summary

Model: openai/gpt-oss-20b
Target layer: 2
Captured region: target_layer_experts_only
Prompts: LMSYS Chat, English-filtered prompt plan
Accepted decode tokens: 600,000
Prompts touched: 9,375
Experts per token: 4
Total expert windows: 2,400,000
Unique experts observed: 32 / 32
Trace sample rate: 10 MSPS
Samples per capture: 100,000
Capture duration: 10 ms
Trace dtype: float16
Trace shape: (100000,)
Approximate trace payload size: 120.08 GB

Layout

traces/shard_XXXXXX/step_YYYYYY.npy      # float16 power trace, shape (100000,)
records/shard_XXXXXX/step_YYYYYY.json    # per-token metadata and expert window bounds
timelines/shard_XXXXXX/...               # periodic timing/timeline artifacts
prompt_plan.jsonl                        # deterministic prompt plan used by the run
capture_meta.json                        # capture summary and expert counts
run_state.json                           # final committed run state
layer_windows.rebuilt.jsonl              # flattened expert-window index from audit

Each record contains:

global_step
prompt_plan_index
token_in_prompt
hit_experts
trace_file
trace_stats
layer_windows, with expert_idx, sample_start, sample_end, and millisecond boundaries
scope configuration including adc_freq_effective, adc_decimate, and adc_samples

Health/audit status

Final audit status:

committed records: 600,000
missing trace files: 0
trace dtypes: float16: 600000
trace shapes: (100000,): 600000
loaded trace sample count during audit: 1,000
audit warnings: 0
audit errors: 0

A separate visual spot check sampled early, middle, and late captures and confirmed that the annotated expert windows are inside the traces and visually aligned.

Notes

The analog chain is AC-coupled/high-pass and the installed current clamp is outside its specified flat bandwidth for MHz-scale components. Treat absolute current calibration cautiously; the main intended use is comparative classification/analysis of expert-conditioned trace structure.

Downloads last month: 31

Total file size:

44 MB