The dataset viewer is not available for this split.
Error code: StreamingRowsError
Exception: CastError
Message: Couldn't cast
array: string
model: string
device: string
batch: int64
vectors: int64
hidden: int64
layers: int64
positions: int64
mismatches: int64
path: string
mid_range_outputs: int64
tokens: int64
tok_per_s: double
macs_per_s: int64
to
{'array': Value('string'), 'model': Value('string'), 'device': Value('string'), 'batch': Value('int64'), 'hidden': Value('int64'), 'layers': Value('int64'), 'positions': Value('int64'), 'mismatches': Value('int64'), 'mid_range_outputs': Value('int64'), 'tokens': Value('int64'), 'tok_per_s': Value('float64')}
because column names don't match
Traceback: Traceback (most recent call last):
File "/src/services/worker/src/worker/utils.py", line 149, 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 129, 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 544, 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 3052, in __iter__
for key, example in ex_iterable:
^^^^^^^^^^^
File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 2559, in __iter__
for key, pa_table in self._iter_arrow():
~~~~~~~~~~~~~~~~^^
File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 2584, 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 558, in _iter_arrow
for key, pa_table in iterator:
^^^^^^^^
File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 440, 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 361, 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 2381, in table_cast
return cast_table_to_schema(table, schema)
File "/usr/local/lib/python3.14/site-packages/datasets/table.py", line 2309, in cast_table_to_schema
raise CastError(
...<3 lines>...
)
datasets.table.CastError: Couldn't cast
array: string
model: string
device: string
batch: int64
vectors: int64
hidden: int64
layers: int64
positions: int64
mismatches: int64
path: string
mid_range_outputs: int64
tokens: int64
tok_per_s: double
macs_per_s: int64
to
{'array': Value('string'), 'model': Value('string'), 'device': Value('string'), 'batch': Value('int64'), 'hidden': Value('int64'), 'layers': Value('int64'), 'positions': Value('int64'), 'mismatches': Value('int64'), 'mid_range_outputs': Value('int64'), 'tokens': Value('int64'), 'tok_per_s': Value('float64')}
because column names don't matchNeed help to make the dataset viewer work? Make sure to review how to configure the dataset viewer, and open a discussion for direct support.
qwenTPU: run logs (no model weights)
qwenTPU compiles a weight-stationary integer matrix-multiply array into the layers of a stock Qwen3.5 (0.8B, 2B or 4B). Output is integer-exact against NumPy. Code and video: https://github.com/Bobliuuu/qwentpu
No weights are released here. Base models are the unmodified Qwen/Qwen3.5-0.8B, -2B, -4B (Apache-2.0, Alibaba); the array is compiled in memory by the code in the GitHub repository (qwentpu/merge.py). This is a dataset-style card for the replay logs, speed logs and the explainer video.
Contents
| path | what |
|---|---|
results/<device>/*.jsonl |
replay of token positions against NumPy per device and array |
results/hw.md |
per-device summary (arrays, positions, mismatches, suite runs) |
results/a100/*.jsonl |
A100 speed logs (whole network, TPU layers only, CUDA graph) and v2 vectorized runs |
results/readme_data.json |
the aggregate used by the README |
qwentpu_explainer.mp4 |
explainer video with voice-over |
Numbers (from the files above)
- 1,361,280 token positions replayed on 5 machines (A100, A10G, L4, T4, x86 Xeon), 0 mismatches.
- A100, Qwen3.5-0.8B, 8x8 u3s3: whole network 6,629 tok/s; TPU layers only 74,462 tok/s; with one CUDA graph per trace about 340,925 tok/s (mean of 3 runs x 2 traces).
- v2 (program in the prompt) random-program exactness runs: 196 of 196 exact.
- T4: 9 of 12 suite runs clean (16 GB memory limit); its replay is exact.
Limits: not faster than QwenPU per operation; v1 is host-driven; teacher-forced numbers are prefill throughput.
Video: https://github.com/Bobliuuu/qwentpu/blob/main/video/qwentpu_explainer.mp4
- Downloads last month
- 85