Dataset Viewer
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
model: string
by_group_size: struct<32: struct<vision: struct<modules_total: int64, modules_quantizable: int64, modules_skipped: (... 2142 chars omitted)
child 0, 32: struct<vision: struct<modules_total: int64, modules_quantizable: int64, modules_skipped: int64, para (... 639 chars omitted)
child 0, vision: struct<modules_total: int64, modules_quantizable: int64, modules_skipped: int64, params_total: int64 (... 71 chars omitted)
child 0, modules_total: int64
child 1, modules_quantizable: int64
child 2, modules_skipped: int64
child 3, params_total: int64
child 4, params_quantizable: int64
child 5, param_coverage: double
child 6, group_size: int64
child 1, projector: struct<modules_total: int64, modules_quantizable: int64, modules_skipped: int64, params_total: int64 (... 71 chars omitted)
child 0, modules_total: int64
child 1, modules_quantizable: int64
child 2, modules_skipped: int64
child 3, params_total: int64
child 4, params_quantizable: int64
child 5, param_coverage: double
child 6, group_size: int64
child 2, backbone: struct<modules_total: int64, modules_quantizable: int64, modules_skipped: int64, params_total: int64 (... 71 chars omitted)
child 0, modules_total: int64
child 1, modules_quantizable: int64
child 2, modules_skipped: int64
child 3, params_total: int64
...
int64
child 5, param_coverage: double
child 6, group_size: int64
child 3, embed_head: struct<modules_total: int64, modules_quantizable: int64, modules_skipped: int64, params_total: int64 (... 71 chars omitted)
child 0, modules_total: int64
child 1, modules_quantizable: int64
child 2, modules_skipped: int64
child 3, params_total: int64
child 4, params_quantizable: int64
child 5, param_coverage: double
child 6, group_size: int64
meta: struct<model: string, precision: string, n: int64, seed: int64, max_image_tokens: int64>
child 0, model: string
child 1, precision: string
child 2, n: int64
child 3, seed: int64
child 4, max_image_tokens: int64
results: list<item: struct<benchmark: string, n: int64, seed: int64, max_image_tokens: int64, max_num_patches (... 187 chars omitted)
child 0, item: struct<benchmark: string, n: int64, seed: int64, max_image_tokens: int64, max_num_patches: int64, ac (... 175 chars omitted)
child 0, benchmark: string
child 1, n: int64
child 2, seed: int64
child 3, max_image_tokens: int64
child 4, max_num_patches: int64
child 5, accuracy: double
child 6, accuracy_given_parsed: double
child 7, parse_failure_rate: double
child 8, n_parsed: int64
child 9, published: double
child 10, wall_seconds: double
child 11, parse_failure_samples: list<item: string>
child 0, item: string
to
{'meta': {'model': Value('string'), 'precision': Value('string'), 'n': Value('int64'), 'seed': Value('int64'), 'max_image_tokens': Value('int64')}, 'results': List({'benchmark': Value('string'), 'n': Value('int64'), 'seed': Value('int64'), 'max_image_tokens': Value('int64'), 'max_num_patches': Value('int64'), 'accuracy': Value('float64'), 'accuracy_given_parsed': Value('float64'), 'parse_failure_rate': Value('float64'), 'n_parsed': Value('int64'), 'published': Value('float64'), 'wall_seconds': Value('float64'), 'parse_failure_samples': List(Value('string'))})}
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 483, 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 2840, in __iter__
for key, example in ex_iterable:
^^^^^^^^^^^
File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 2373, in __iter__
for key, pa_table in self._iter_arrow():
~~~~~~~~~~~~~~~~^^
File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 2398, 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 2378, in table_cast
return cast_table_to_schema(table, schema)
File "/usr/local/lib/python3.14/site-packages/datasets/table.py", line 2306, in cast_table_to_schema
raise CastError(
...<3 lines>...
)
datasets.table.CastError: Couldn't cast
model: string
by_group_size: struct<32: struct<vision: struct<modules_total: int64, modules_quantizable: int64, modules_skipped: (... 2142 chars omitted)
child 0, 32: struct<vision: struct<modules_total: int64, modules_quantizable: int64, modules_skipped: int64, para (... 639 chars omitted)
child 0, vision: struct<modules_total: int64, modules_quantizable: int64, modules_skipped: int64, params_total: int64 (... 71 chars omitted)
child 0, modules_total: int64
child 1, modules_quantizable: int64
child 2, modules_skipped: int64
child 3, params_total: int64
child 4, params_quantizable: int64
child 5, param_coverage: double
child 6, group_size: int64
child 1, projector: struct<modules_total: int64, modules_quantizable: int64, modules_skipped: int64, params_total: int64 (... 71 chars omitted)
child 0, modules_total: int64
child 1, modules_quantizable: int64
child 2, modules_skipped: int64
child 3, params_total: int64
child 4, params_quantizable: int64
child 5, param_coverage: double
child 6, group_size: int64
child 2, backbone: struct<modules_total: int64, modules_quantizable: int64, modules_skipped: int64, params_total: int64 (... 71 chars omitted)
child 0, modules_total: int64
child 1, modules_quantizable: int64
child 2, modules_skipped: int64
child 3, params_total: int64
...
int64
child 5, param_coverage: double
child 6, group_size: int64
child 3, embed_head: struct<modules_total: int64, modules_quantizable: int64, modules_skipped: int64, params_total: int64 (... 71 chars omitted)
child 0, modules_total: int64
child 1, modules_quantizable: int64
child 2, modules_skipped: int64
child 3, params_total: int64
child 4, params_quantizable: int64
child 5, param_coverage: double
child 6, group_size: int64
meta: struct<model: string, precision: string, n: int64, seed: int64, max_image_tokens: int64>
child 0, model: string
child 1, precision: string
child 2, n: int64
child 3, seed: int64
child 4, max_image_tokens: int64
results: list<item: struct<benchmark: string, n: int64, seed: int64, max_image_tokens: int64, max_num_patches (... 187 chars omitted)
child 0, item: struct<benchmark: string, n: int64, seed: int64, max_image_tokens: int64, max_num_patches: int64, ac (... 175 chars omitted)
child 0, benchmark: string
child 1, n: int64
child 2, seed: int64
child 3, max_image_tokens: int64
child 4, max_num_patches: int64
child 5, accuracy: double
child 6, accuracy_given_parsed: double
child 7, parse_failure_rate: double
child 8, n_parsed: int64
child 9, published: double
child 10, wall_seconds: double
child 11, parse_failure_samples: list<item: string>
child 0, item: string
to
{'meta': {'model': Value('string'), 'precision': Value('string'), 'n': Value('int64'), 'seed': Value('int64'), 'max_image_tokens': Value('int64')}, 'results': List({'benchmark': Value('string'), 'n': Value('int64'), 'seed': Value('int64'), 'max_image_tokens': Value('int64'), 'max_num_patches': Value('int64'), 'accuracy': Value('float64'), 'accuracy_given_parsed': Value('float64'), 'parse_failure_rate': Value('float64'), 'n_parsed': Value('int64'), 'published': Value('float64'), 'wall_seconds': Value('float64'), 'parse_failure_samples': List(Value('string'))})}
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.
lfm2vl-lab — raw evaluation results
Every number behind lfm2vl-lab, a
component-wise quantization and edge-inference study of
LiquidAI/LFM2.5-VL-3B on Apple Silicon
(M4 Pro, 24 GB).
Published separately from the code so the numbers stay citable and diffable on their own. Weights in that study are ephemeral by design — converted, evaluated, recorded, deleted. These JSON files are what survives.
Files
| file | what it holds |
|---|---|
coverage_audit_gs64.json |
per-module quantizability audit at group size 64 |
coverage_audit_by_group_size.json |
the same at group size 32 / 64 / 128 |
uniform4_coverage_verified.json |
coverage read back from saved weights |
published_quant_survey.json |
56 published LFM2-VL/LFM2.5-VL quant repos, checked for vision-tower quantization |
token_budget_sweep.json |
accuracy and latency vs max_image_tokens, 3 tasks × 4 budgets |
baseline_bf16_n200_t512.json |
bf16 baseline vs Liquid's published figures |
sweep_component_sensitivity.json |
17 quantization configs × 5 benchmarks |
head_to_head_n300.json |
5 decision-relevant configs at n=300 with 95% CIs |
Headline findings
The vision token budget matters more than quantization. ScreenSpot-v2 moves from
53.0 to 79.0 by changing max_image_tokens from the default 256 to 512 — and falls
back to 70.0 at 1024. Non-monotonic, and the optimum is task-dependent (512 for screens
and documents, 256 for natural images).
Component sensitivity is flat. Across 17 configs, nothing measurably degrades at 6 or 8 bits on any component. The only clear loser is uniform 4-bit, the one config pushing the vision tower to 4 bits — lower on 5 of 5 benchmarks.
A third of the vision tower cannot be quantized. SigLIP2 shape-optimized has
intermediate_size = 4304 (16 × 269, 269 prime), so all 27 mlp.fc2 modules are
skipped at every standard group size — silently, with no warning.
No published quant touches the vision tower. Of 56 repos surveyed, including
LiquidAI's own MLX releases, none has a vision_tower quantization entry.
Caveats
n is 100–300 per benchmark, so 95% CIs run ±4 to ±5 points. Differences smaller than
that are not claims. Single machine, single seed (0), mlx-vlm==0.6.13.
- Downloads last month
- -