Cloud Atlas
Collection
A collection of model atlases for mechanistic interpretation I've created. • 9 items • Updated
Datasets:
Tasks:
Feature Extraction
Languages:
English
Size:
100K<n<1M
Tags:
SAE
mechinterp
mechanistic-interpretability
sparse-autoencoders
monosemanticity
feature-extraction
License:
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: TypeError
Message: Couldn't cast array of type
struct<top: double, mean: double, top_idx: int64>
to
{'code_bucket': Value('string'), 'code_entangled': Value('int64'), 'code_selective': Value('int64')}
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 2303, in cast_table_to_schema
cast_array_to_feature(
~~~~~~~~~~~~~~~~~~~~~^
table[name] if name in table_column_names else pa.array([None] * len(table), type=schema.field(name).type),
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
feature,
^^^^^^^^
)
^
File "/usr/local/lib/python3.14/site-packages/datasets/table.py", line 1852, in wrapper
return pa.chunked_array([func(chunk, *args, **kwargs) for chunk in array.chunks])
~~~~^^^^^^^^^^^^^^^^^^^^^^^^
File "/usr/local/lib/python3.14/site-packages/datasets/table.py", line 2059, in cast_array_to_feature
_c(array.field(name) if name in array_fields else null_array, subfeature)
~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/usr/local/lib/python3.14/site-packages/datasets/table.py", line 1854, in wrapper
return func(array, *args, **kwargs)
File "/usr/local/lib/python3.14/site-packages/datasets/table.py", line 2149, in cast_array_to_feature
raise TypeError(f"Couldn't cast array of type\n{_short_str(array.type)}\nto\n{_short_str(feature)}")
TypeError: Couldn't cast array of type
struct<top: double, mean: double, top_idx: int64>
to
{'code_bucket': Value('string'), 'code_entangled': Value('int64'), 'code_selective': Value('int64')}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.
VibeThinker-1.5B Brain Atlas
This is an internal-mechanics atlas for the 1.5B parameter VibeThinker model. The goal was not to benchmark end-task accuracy, but to map what the network is actually doing with its parameters: where it computes, where it stores behaviorally relevant structure, and which late-layer directions are safe to touch.
What was run
- Activation census over 9,523 prompts spanning compliance, reasoning, code, math, multilingual, and refusal-style questions.
- Per-layer feature taxonomy for
mlp,gate,up, and attention heads. - OV-circuit spectral analysis per head (
W_V @ W_O). - Sub-Zero surgery pass on every layer, with a capability fence across
code,math,reasoning,factual, andmultilingualdomains. - Pipeline was run on a CPU-only environment.
Key geometry
| Property | Value |
|---|---|
| Layers | 28 |
| d_model | 1536 |
| d_mlp | 8960 |
| Attention heads | 12 |
| KV heads | 2 |
| Head dim | 128 |
| Sacred (deep Sub-Zero) layers | 18–27 |
What the numbers suggest
The model is not a lookup table
OV-circuit spectral concentration averages 0.049, with effective rank around 55. That is a distributed signature, not a sparse “copy-paste” attention pattern. Attention heads appear to be doing weighted computation across many directions, not memorizing specific token-to-token jumps.
Feature activation is broad
The feature taxonomy is dominated by partial_shared and broadly_shared classes, with a smaller non_activated tail and very few all_shared features. Most dimensions responded to many prompts rather than one hyper-specific trigger.
Late layers are load-bearing
Sub-Zero finds structured singular-value subspace only in layers 18–27, which is 36% of the network depth. The first half of the model looks like wide preprocessing; the second half does the structured transformation.
Surgical fragility is the main caveat
The capability fence keeps about 74.5% of tested axes, but the rejected ones hit hard:
- Layer 18
up_projaxis 0 does 0.81 damage to code generation. - Layer 18
up_projaxis 0 also scores the highest math, reasoning, and multilingual damage. - Several
down_projandgate_projaxes in the early sacred layers fail the fence.
Interpretation: the 1.5B late-layer subspace is doing a lot of work per direction. It has less redundancy than the larger variant, so removing a top singular value tends to break more than one capability at once.
Classifier stability dips in the middle
Sub-Zero classifier accuracy drops to 0.75–0.83 around layers 13–17, then recovers in the late sacred layers. That mid-network region is messier or more entangled than the clean late-layer representation.
Bottom line
VibeThinker-1.5B behaves like a compact reasoning model: distributed attention, broad-feature MLPs, and a deep-but-narrow sacred region where a small number of directions carry most of the task load. It is interpretable, but not easy to edit safely because its late layers are not highly redundant.
- Downloads last month
- 869