Hugging Face's logo

Datasets:
siddharthmb
/
2026.TA.attribution_big_graphs

ArXiv:
Tags:
interpretability
circuit-tracing
attribution-graphs
transcoder-adapters
gemma-2-2b
Dataset card Files
xet
 Community
Dataset Viewer
Duplicate
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:    ArrowInvalid
Message:      JSON parse error: Missing a name for object member. in row 0
Traceback:    Traceback (most recent call last):
                File "/usr/local/lib/python3.14/site-packages/datasets/packaged_modules/json/json.py", line 324, in _generate_tables
                  df = pandas_read_json(f)
                File "/usr/local/lib/python3.14/site-packages/datasets/packaged_modules/json/json.py", line 38, in pandas_read_json
                  return pd.read_json(path_or_buf, **kwargs)
                         ~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^
                File "/usr/local/lib/python3.14/site-packages/pandas/io/json/_json.py", line 815, in read_json
                  return json_reader.read()
                         ~~~~~~~~~~~~~~~~^^
                File "/usr/local/lib/python3.14/site-packages/pandas/io/json/_json.py", line 1014, in read
                  obj = self._get_object_parser(self.data)
                File "/usr/local/lib/python3.14/site-packages/pandas/io/json/_json.py", line 1040, in _get_object_parser
                  obj = FrameParser(json, **kwargs).parse()
                File "/usr/local/lib/python3.14/site-packages/pandas/io/json/_json.py", line 1176, in parse
                  self._parse()
                  ~~~~~~~~~~~^^
                File "/usr/local/lib/python3.14/site-packages/pandas/io/json/_json.py", line 1391, in _parse
                  self.obj = DataFrame(
                             ~~~~~~~~~^
                      ujson_loads(json, precise_float=self.precise_float), dtype=None
                      ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
                  )
                  ^
                File "/usr/local/lib/python3.14/site-packages/pandas/core/frame.py", line 782, in __init__
                  mgr = dict_to_mgr(data, index, columns, dtype=dtype, copy=copy, typ=manager)
                File "/usr/local/lib/python3.14/site-packages/pandas/core/internals/construction.py", line 503, in dict_to_mgr
                  return arrays_to_mgr(arrays, columns, index, dtype=dtype, typ=typ, consolidate=copy)
                File "/usr/local/lib/python3.14/site-packages/pandas/core/internals/construction.py", line 114, in arrays_to_mgr
                  index = _extract_index(arrays)
                File "/usr/local/lib/python3.14/site-packages/pandas/core/internals/construction.py", line 677, in _extract_index
                  raise ValueError("All arrays must be of the same length")
              ValueError: All arrays must be of the same length
              
              During handling of the above exception, another exception occurred:
              
              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 2951, in __iter__
                  for key, example in ex_iterable:
                                      ^^^^^^^^^^^
                File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 2461, in __iter__
                  for key, pa_table in self._iter_arrow():
                                       ~~~~~~~~~~~~~~~~^^
                File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 2486, 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 547, in _iter_arrow
                  for key, pa_table in iterator:
                                       ^^^^^^^^
                File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 430, 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 327, in _generate_tables
                  raise e
                File "/usr/local/lib/python3.14/site-packages/datasets/packaged_modules/json/json.py", line 290, in _generate_tables
                  pa_table = paj.read_json(
                      io.BytesIO(batch), read_options=paj.ReadOptions(block_size=block_size)
                  )
                File "pyarrow/_json.pyx", line 364, in pyarrow._json.read_json
                File "pyarrow/error.pxi", line 155, in pyarrow.lib.pyarrow_internal_check_status
                File "pyarrow/error.pxi", line 92, in pyarrow.lib.check_status
                  raise convert_status(status)
              pyarrow.lib.ArrowInvalid: JSON parse error: Missing a name for object member. in row 0

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.

2026.TA.attribution_big_graphs

Base-vs-adapter attribution graphs for the combined_big run, built with python -m analysis.attribution.graphs.build_graphs --graph_kind fused with graph kind(s) fused; the graphs sit at the repo root.

Combined base+adapter attribution graphs for the combined_big run: ONE attribution graph per prompt over a single forward pass of the model named by the backbone below. Each MLP is replaced by T_base(x) + T_adapter(x) + Err, so every graph holds GemmaScope base features (hexagon ⬢), transcoder-adapter features (circle ●) and reconstruction-error nodes (triangle ▲) under one shared set of logit nodes; base<->adapter edges and edges into the logits are real attribution. This differs from the overlay graphs (--graph_kind overlay), which glue two separately computed graphs that share only embedding nodes.

Backbone: base. The combined replacement runs on the plain google/gemma-2-2b backbone, whose forward pass collapses to the base model, so the logits and continuations are the BASE model's; read adapter features here as a decomposition of the base model's computation, not as the adapter model's circuit. (The manifest predates the backbone key; all such runs used the base backbone.)

A 2026-06/07 combined (fused) graph set of experiments/base_vs_adapter_circuit_trace: one attribution graph per prompt over the adapter model with GemmaScope base features and adapter features as nodes; the dtk20 and ms100000 suffixes name the feature scan the examples came from. The study's graphs/ in the repository holds the subset its writeup cites.

Uploaded unchanged from the cluster's scratch volume by the disk audit of 2026-10-11 (claude_updates/memory/reports/disk_audit.md) so the scratch copy could be deleted. A historical set: the refusal-tracing, fused-graph and input-shift writeups cite the sets they name, and the headline rerun of 2026-10 (siddharthmb/2026.TA.gemma2_2b_headline_2026-10d_tc8192_decb_l1w0.001_auxk512_fp32_tarbb_lb2_ln1_sl17824146; the sets siddharthmb/2026.TA.tc8192_refusal_first10_headline_2026-10d_graphs, siddharthmb/2026.TA.tc8192_compliance_first10_headline_2026-10d_graphs) supersedes the adapter-side claims (docs/decisions.md §9, §11).

How to use 

uv run --extra viz python -m analysis.serve siddharthmb/2026.TA.attribution_big_graphs --port 8053

The graphs sit at the repo root, so the bare repo id is the graph dir (no :subdir). The snapshot is downloaded and cached on first run (HF_HOME); node-click feature examples stream from the feature collections linked below.

How it was produced

Produced by python -m analysis.attribution.graphs.build_graphs --graph_kind fused from transcoder-adapters.

Command (from the run's Slurm log; the module name is updated to the current entry point, flags unchanged):

uv run --no-sync python -m analysis.attribution.graphs.build_graphs --graph_kind fused --adapter_checkpoint siddharthmb/2026.TA.gemma2_2b_tc8192_decb_l1w0.001_tarbb_lb2.0_ln1_dr20000_lr8e-04_bs4_sl14793860 --base_model google/gemma-2-2b --prompts analysis/attribution/prompts/interesting_small --prompt_format chat --gemmascope_width width_16k --gemmascope_l0 average_l0_76 --base_feature_data_path siddharthmb/2026.TA.features_gemma-2-2b_gemmascope_width_16k_average_l0_76_ms5000_ml1024_tk10_h8f13c4bce9be --adapter_feature_data_path siddharthmb/2026.TA.features_2026.TA.gemma2_2b_tc8192_decb_l1w0.001_tarbb_lb2.0_ln1_dr20000_lr_hdcac93c56a93 --max_feature_nodes 8192 --batch_size 4 --max_n_logits 10 --max_error_nodes 64 --run_name combined_big --output_dir /nlp/scr/siddharth/transcoder-adapters/attribution_big

Re-star the jailbreak/compliance prompts in the dropdown after any rebuild (idempotent, auto-detects the set from the baked adapter continuations):

uv run python -m analysis.evals.classify_adapter_compliance \
  --graph_dir /nlp/scr/siddharth/transcoder-adapters/attribution_big --star

Uploaded with:

uv run python -m analysis.attribution.hub --graph_dir /nlp/scr/siddharth/transcoder-adapters/attribution_big

Details

Contents

path files size
graphs (repo root) 12 809.5 MB
top-level files 2 10.4 KB

Inputs

field value
adapter checkpoint siddharthmb/2026.TA.gemma2_2b_tc8192_decb_l1w0.001_tarbb_lb2.0_ln1_dr20000_lr8e-04_bs4_sl14793860
base model google/gemma-2-2b
adapter feature examples siddharthmb/2026.TA.features_2026.TA.gemma2_2b_tc8192_decb_l1w0.001_tarbb_lb2.0_ln1_dr20000_lr_hdcac93c56a93
base feature examples siddharthmb/2026.TA.features_gemma-2-2b_gemmascope_width_16k_average_l0_76_ms5000_ml1024_tk10_h8f13c4bce9be
prompt set ``
prompt format ``
backbone base (manifest predates the key)
n_base_features 16384 (combined feature index < this is a base feature)
max_error_nodes 64
prompts traced 12 of 12 succeeded

Results

Graphs

graph base features adapter features error nodes
combined_big__bomb_refusal_help_I__hafcdd5f8d6c1 n/a n/a n/a
combined_big__capital_colesseum__he6b0cdbc003d n/a n/a n/a
combined_big__capital_colesseum_mispelling__h6c3a39bbc532 n/a n/a n/a
combined_big__capital_paris__h4f93ff3e59c4 n/a n/a n/a
combined_big__capital_paris_eos__h0778de7116be n/a n/a n/a
combined_big__cookies_explosives_however__h7d5edf13528e n/a n/a n/a
combined_big__favorite_drink__h52755272bf00 n/a n/a n/a
combined_big__hacking_refusal_assist__h56595634e1ac n/a n/a n/a
combined_big__math_forty_two_first__h1426d9baeaab n/a n/a n/a
combined_big__math_forty_two_second__h4bf177829657 n/a n/a n/a
combined_big__opposite_cold__h407abc63d121 n/a n/a n/a
combined_big__poison_refusal_help__h04f174d68262 n/a n/a n/a

Node counts are n/a because this run's graph-metadata.json predates recording them; each graph's metadata.comparison.node_counts has them.

Links

Cluster paths

  • graph dir: /nlp/scr/siddharth/transcoder-adapters/attribution_big
  • manifest: /nlp/scr/siddharth/transcoder-adapters/attribution_big/combined_attribution_manifest.json (also in this repo)
  • build log: /juice2/u/siddharth/transcoder-adapters/logs/combined_attribution/20260625_143936_15996691.out
Downloads last month
65
Total file size:
849 MB

Paper for siddharthmb/2026.TA.attribution_big_graphs