Glimmer Scope
Collection
Sparse autoencoders on nine layers of Muse Glimmer 30B. Report: doi.org/10.5281/zenodo.23952447 · Atlas: scope.kroonen.ai • 4 items • Updated • 1
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
prompts: list<item: string>
child 0, item: string
baseline: list<item: string>
child 0, item: string
examples: list<item: struct<layer: int64, id: int64, label: string, does: string, strength: double, replies: l (... 19 chars omitted)
child 0, item: struct<layer: int64, id: int64, label: string, does: string, strength: double, replies: list<item: s (... 7 chars omitted)
child 0, layer: int64
child 1, id: int64
child 2, label: string
child 3, does: string
child 4, strength: double
child 5, replies: list<item: string>
child 0, item: string
presets: list<item: struct<name: string, mix: struct<pirate: double, british: double, noir: double, paris: do (... 134 chars omitted)
child 0, item: struct<name: string, mix: struct<pirate: double, british: double, noir: double, paris: double, santa (... 122 chars omitted)
child 0, name: string
child 1, mix: struct<pirate: double, british: double, noir: double, paris: double, santa: double, ocean: double, s (... 95 chars omitted)
child 0, pirate: double
child 1, british: double
child 2, noir: double
child 3, paris: double
child 4, santa: double
child 5, ocean: double
child 6, stoic: double
child 7, grandparent: double
child 8, fairies: double
child 9, chess: double
child 10, winter: double
child 11, ink: double
median_norm: struct<24: double, 36: double>
child 0, 24: double
child 1, 36: double
dials: list<item: struct<key: string, label: string, group: string, layer: int64, default: double, descript (... 106 chars omitted)
child 0, item: struct<key: string, label: string, group: string, layer: int64, default: double, description: string (... 94 chars omitted)
child 0, key: string
child 1, label: string
child 2, group: string
child 3, layer: int64
child 4, default: double
child 5, description: string
child 6, kind: string
child 7, feature: int64
child 8, clean: bool
child 9, judge_said: string
child 10, fires_on: list<item: string>
child 0, item: string
to
{'median_norm': {'24': Value('float64'), '36': Value('float64')}, 'dials': List({'key': Value('string'), 'label': Value('string'), 'group': Value('string'), 'layer': Value('int64'), 'default': Value('float64'), 'description': Value('string'), 'kind': Value('string'), 'feature': Value('int64'), 'clean': Value('bool'), 'judge_said': Value('string'), 'fires_on': List(Value('string'))}), 'presets': List({'name': Value('string'), 'mix': {'pirate': Value('float64'), 'british': Value('float64'), 'noir': Value('float64'), 'paris': Value('float64'), 'santa': Value('float64'), 'ocean': Value('float64'), 'stoic': Value('float64'), 'grandparent': Value('float64'), 'fairies': Value('float64'), 'chess': Value('float64'), 'winter': Value('float64'), 'ink': 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
prompts: list<item: string>
child 0, item: string
baseline: list<item: string>
child 0, item: string
examples: list<item: struct<layer: int64, id: int64, label: string, does: string, strength: double, replies: l (... 19 chars omitted)
child 0, item: struct<layer: int64, id: int64, label: string, does: string, strength: double, replies: list<item: s (... 7 chars omitted)
child 0, layer: int64
child 1, id: int64
child 2, label: string
child 3, does: string
child 4, strength: double
child 5, replies: list<item: string>
child 0, item: string
presets: list<item: struct<name: string, mix: struct<pirate: double, british: double, noir: double, paris: do (... 134 chars omitted)
child 0, item: struct<name: string, mix: struct<pirate: double, british: double, noir: double, paris: double, santa (... 122 chars omitted)
child 0, name: string
child 1, mix: struct<pirate: double, british: double, noir: double, paris: double, santa: double, ocean: double, s (... 95 chars omitted)
child 0, pirate: double
child 1, british: double
child 2, noir: double
child 3, paris: double
child 4, santa: double
child 5, ocean: double
child 6, stoic: double
child 7, grandparent: double
child 8, fairies: double
child 9, chess: double
child 10, winter: double
child 11, ink: double
median_norm: struct<24: double, 36: double>
child 0, 24: double
child 1, 36: double
dials: list<item: struct<key: string, label: string, group: string, layer: int64, default: double, descript (... 106 chars omitted)
child 0, item: struct<key: string, label: string, group: string, layer: int64, default: double, description: string (... 94 chars omitted)
child 0, key: string
child 1, label: string
child 2, group: string
child 3, layer: int64
child 4, default: double
child 5, description: string
child 6, kind: string
child 7, feature: int64
child 8, clean: bool
child 9, judge_said: string
child 10, fires_on: list<item: string>
child 0, item: string
to
{'median_norm': {'24': Value('float64'), '36': Value('float64')}, 'dials': List({'key': Value('string'), 'label': Value('string'), 'group': Value('string'), 'layer': Value('int64'), 'default': Value('float64'), 'description': Value('string'), 'kind': Value('string'), 'feature': Value('int64'), 'clean': Value('bool'), 'judge_said': Value('string'), 'fires_on': List(Value('string'))}), 'presets': List({'name': Value('string'), 'mix': {'pirate': Value('float64'), 'british': Value('float64'), 'noir': Value('float64'), 'paris': Value('float64'), 'santa': Value('float64'), 'ocean': Value('float64'), 'stoic': Value('float64'), 'grandparent': Value('float64'), 'fairies': Value('float64'), 'chess': Value('float64'), 'winter': Value('float64'), 'ink': 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.
Glimmer Scope Space data
The data behind the Glimmer Scope Space: the 20 dials (unit steering directions from the Glimmer Scope SAEs on Muse Glimmer 30B, plus what each one does), the recorded steering examples, and the atlas images.
| File | What it is |
|---|---|
dials.safetensors |
one unit direction per dial, 6,656 floats each |
dials.json |
each dial's label, layer, default strength, description, and the layers' median residual norms |
examples.json |
steering-test replies to four fixed questions, for nine features |
assets/ |
the Ocean feature's thread through the nine layers, light and dark |
Kroonen AI
- Downloads last month
- 8