Datasets:

siddharthmb
/

2026.TA.tc8192_strict_refusal_graphs

Tags:

Dataset Viewer

Split (1)

train

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
graphs: list<item: struct<slug: string, scan: string, transcoder_list: list<item: null>, prompt_tokens: list (... 443 chars omitted)
  child 0, item: struct<slug: string, scan: string, transcoder_list: list<item: null>, prompt_tokens: list<item: stri (... 431 chars omitted)
      child 0, slug: string
      child 1, scan: string
      child 2, transcoder_list: list<item: null>
          child 0, item: null
      child 3, prompt_tokens: list<item: string>
          child 0, item: string
      child 4, prompt: string
      child 5, node_threshold: double
      child 6, schema_version: int64
      child 7, comparison: struct<base_slug: string, adapter_slug: string, base_scan: string, adapter_scan: string, base_featur (... 250 chars omitted)
          child 0, base_slug: string
          child 1, adapter_slug: string
          child 2, base_scan: string
          child 3, adapter_scan: string
          child 4, base_feature_scan: string
          child 5, adapter_feature_scan: string
          child 6, node_shapes: struct<base: string, adapter: string, shared: string>
              child 0, base: string
              child 1, adapter: string
              child 2, shared: string
          child 7, link_filter: struct<hide_direct_embedding_logit_links: bool, skipped_direct_embedding_logit_links: struct<base: i (... 22 chars omitted)
              child 0, hide_direct_embedding_logit_links: bool
              child 1, skipped_direct_embedding_logit_links: struct<base: int64, adapte
...
ring
      child 15, harm_040: string
      child 16, harm_045: string
      child 17, harm_055: string
  child 1, adapter: struct<harm_000: string, harm_002: string, harm_004: string, harm_006: string, harm_010: string, har (... 230 chars omitted)
      child 0, harm_000: string
      child 1, harm_002: string
      child 2, harm_004: string
      child 3, harm_006: string
      child 4, harm_010: string
      child 5, harm_011: string
      child 6, harm_012: string
      child 7, harm_016: string
      child 8, harm_017: string
      child 9, harm_021: string
      child 10, harm_027: string
      child 11, harm_028: string
      child 12, harm_031: string
      child 13, harm_034: string
      child 14, harm_035: string
      child 15, harm_040: string
      child 16, harm_045: string
      child 17, harm_055: string
prompt_tokenizer_model: string
adapter_checkpoint: string
gemmascope: struct<repo: string, width: string, requested_l0: string, l0_match: string, l0_by_layer: list<item:  (... 39 chars omitted)
  child 0, repo: string
  child 1, width: string
  child 2, requested_l0: string
  child 3, l0_match: string
  child 4, l0_by_layer: list<item: string>
      child 0, item: string
  child 5, n_layers: int64
  child 6, scan: string
run_name: string
base_model: string
overlay_graph_paths: list<item: string>
  child 0, item: string
prompt_format: string
base_graph_dir: string
overlay_graph_dir: string
compact_overlay_graph_paths: list<item: string>
  child 0, item: string
to
{'adapter_checkpoint': Value('string'), 'base_model': Value('string'), 'prompt_tokenizer_model': Value('string'), 'prompts': Value('string'), 'run_name': Value('string'), 'prompt_format': Value('string'), 'output_dir': Value('string'), 'base_graph_dir': Value('string'), 'adapter_graph_dir': Value('string'), 'overlay_graph_dir': Value('string'), 'overlay_graph_paths': List(Value('string')), 'compact_overlay_graph_dir': Value('string'), 'compact_overlay_graph_paths': List(Value('string')), 'gemmascope_config_path': Value('string'), 'gemmascope': {'repo': Value('string'), 'width': Value('string'), 'requested_l0': Value('string'), 'l0_match': Value('string'), 'l0_by_layer': List(Value('string')), 'n_layers': Value('int64'), 'scan': Value('string')}, 'adapter_features': {'feature_data_path': Value('string'), 'feature_output_dir': Value('null'), 'scan': Value('string'), 'features_dir': Value('null')}, 'base_features': {'feature_data_path': Value('string'), 'scan': Value('string'), 'features_dir': Value('null')}, 'overlay_options': {'hide_direct_embedding_logit_links': Value('bool'), 'compact_max_base_feature_nodes': Value('int64'), 'compact_max_base_error_nodes': Value('int64'), 'compact_max_adapter_feature_nodes': Value('null')}, 'results': {'base': {'harm_000': Value('string'), 'harm_002': Value('string'), 'harm_004': Value('string'), 'harm_006': Value('string'), 'harm_010': Value('string'), 'harm_011': Value('string'), 'harm_012': Value('string'), 'harm_016': Value('string'), 'harm_017': Value('string'), 'harm_021': Value('string'), 'harm_027': Value('string'), 'harm_028': Value('string'), 'harm_031': Value('string'), 'harm_034': Value('string'), 'harm_035': Value('string'), 'harm_040': Value('string'), 'harm_045': Value('string'), 'harm_055': Value('string')}, 'adapter': {'harm_000': Value('string'), 'harm_002': Value('string'), 'harm_004': Value('string'), 'harm_006': Value('string'), 'harm_010': Value('string'), 'harm_011': Value('string'), 'harm_012': Value('string'), 'harm_016': Value('string'), 'harm_017': Value('string'), 'harm_021': Value('string'), 'harm_027': Value('string'), 'harm_028': Value('string'), 'harm_031': Value('string'), 'harm_034': Value('string'), 'harm_035': Value('string'), 'harm_040': Value('string'), 'harm_045': Value('string'), 'harm_055': Value('string')}}, 'later_todo': 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
              graphs: list<item: struct<slug: string, scan: string, transcoder_list: list<item: null>, prompt_tokens: list (... 443 chars omitted)
                child 0, item: struct<slug: string, scan: string, transcoder_list: list<item: null>, prompt_tokens: list<item: stri (... 431 chars omitted)
                    child 0, slug: string
                    child 1, scan: string
                    child 2, transcoder_list: list<item: null>
                        child 0, item: null
                    child 3, prompt_tokens: list<item: string>
                        child 0, item: string
                    child 4, prompt: string
                    child 5, node_threshold: double
                    child 6, schema_version: int64
                    child 7, comparison: struct<base_slug: string, adapter_slug: string, base_scan: string, adapter_scan: string, base_featur (... 250 chars omitted)
                        child 0, base_slug: string
                        child 1, adapter_slug: string
                        child 2, base_scan: string
                        child 3, adapter_scan: string
                        child 4, base_feature_scan: string
                        child 5, adapter_feature_scan: string
                        child 6, node_shapes: struct<base: string, adapter: string, shared: string>
                            child 0, base: string
                            child 1, adapter: string
                            child 2, shared: string
                        child 7, link_filter: struct<hide_direct_embedding_logit_links: bool, skipped_direct_embedding_logit_links: struct<base: i (... 22 chars omitted)
                            child 0, hide_direct_embedding_logit_links: bool
                            child 1, skipped_direct_embedding_logit_links: struct<base: int64, adapte
              ...
              ring
                    child 15, harm_040: string
                    child 16, harm_045: string
                    child 17, harm_055: string
                child 1, adapter: struct<harm_000: string, harm_002: string, harm_004: string, harm_006: string, harm_010: string, har (... 230 chars omitted)
                    child 0, harm_000: string
                    child 1, harm_002: string
                    child 2, harm_004: string
                    child 3, harm_006: string
                    child 4, harm_010: string
                    child 5, harm_011: string
                    child 6, harm_012: string
                    child 7, harm_016: string
                    child 8, harm_017: string
                    child 9, harm_021: string
                    child 10, harm_027: string
                    child 11, harm_028: string
                    child 12, harm_031: string
                    child 13, harm_034: string
                    child 14, harm_035: string
                    child 15, harm_040: string
                    child 16, harm_045: string
                    child 17, harm_055: string
              prompt_tokenizer_model: string
              adapter_checkpoint: string
              gemmascope: struct<repo: string, width: string, requested_l0: string, l0_match: string, l0_by_layer: list<item:  (... 39 chars omitted)
                child 0, repo: string
                child 1, width: string
                child 2, requested_l0: string
                child 3, l0_match: string
                child 4, l0_by_layer: list<item: string>
                    child 0, item: string
                child 5, n_layers: int64
                child 6, scan: string
              run_name: string
              base_model: string
              overlay_graph_paths: list<item: string>
                child 0, item: string
              prompt_format: string
              base_graph_dir: string
              overlay_graph_dir: string
              compact_overlay_graph_paths: list<item: string>
                child 0, item: string
              to
              {'adapter_checkpoint': Value('string'), 'base_model': Value('string'), 'prompt_tokenizer_model': Value('string'), 'prompts': Value('string'), 'run_name': Value('string'), 'prompt_format': Value('string'), 'output_dir': Value('string'), 'base_graph_dir': Value('string'), 'adapter_graph_dir': Value('string'), 'overlay_graph_dir': Value('string'), 'overlay_graph_paths': List(Value('string')), 'compact_overlay_graph_dir': Value('string'), 'compact_overlay_graph_paths': List(Value('string')), 'gemmascope_config_path': Value('string'), 'gemmascope': {'repo': Value('string'), 'width': Value('string'), 'requested_l0': Value('string'), 'l0_match': Value('string'), 'l0_by_layer': List(Value('string')), 'n_layers': Value('int64'), 'scan': Value('string')}, 'adapter_features': {'feature_data_path': Value('string'), 'feature_output_dir': Value('null'), 'scan': Value('string'), 'features_dir': Value('null')}, 'base_features': {'feature_data_path': Value('string'), 'scan': Value('string'), 'features_dir': Value('null')}, 'overlay_options': {'hide_direct_embedding_logit_links': Value('bool'), 'compact_max_base_feature_nodes': Value('int64'), 'compact_max_base_error_nodes': Value('int64'), 'compact_max_adapter_feature_nodes': Value('null')}, 'results': {'base': {'harm_000': Value('string'), 'harm_002': Value('string'), 'harm_004': Value('string'), 'harm_006': Value('string'), 'harm_010': Value('string'), 'harm_011': Value('string'), 'harm_012': Value('string'), 'harm_016': Value('string'), 'harm_017': Value('string'), 'harm_021': Value('string'), 'harm_027': Value('string'), 'harm_028': Value('string'), 'harm_031': Value('string'), 'harm_034': Value('string'), 'harm_035': Value('string'), 'harm_040': Value('string'), 'harm_045': Value('string'), 'harm_055': Value('string')}, 'adapter': {'harm_000': Value('string'), 'harm_002': Value('string'), 'harm_004': Value('string'), 'harm_006': Value('string'), 'harm_010': Value('string'), 'harm_011': Value('string'), 'harm_012': Value('string'), 'harm_016': Value('string'), 'harm_017': Value('string'), 'harm_021': Value('string'), 'harm_027': Value('string'), 'harm_028': Value('string'), 'harm_031': Value('string'), 'harm_034': Value('string'), 'harm_035': Value('string'), 'harm_040': Value('string'), 'harm_045': Value('string'), 'harm_055': Value('string')}}, 'later_todo': List(Value('string'))}
              because column names don't match

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.tc8192_strict_refusal_graphs

Base-vs-adapter circuit-tracing comparison overlay graphs for the tc8192_strict_refusal run. Each overlay graph is one full-replacement attribution graph over a single prompt in which MLP(x) = T_base(x) + T_adapter(x) + Err: base GemmaScope transcoder features (hexagon ⬢), sparse transcoder-adapter features (circle ●) and real reconstruction-error nodes (triangle ▲) live in the same graph, so the two models' circuits can be read position-by-position on identical tokens.

What this run is

The original chat-template refusal overlays on the deployed tc8192 adapter (18 strict-refusal prompts) — the run served on port 8050 in sh/visualize graphs/visualize_base_adapter 07-29-2026.sh.

⚠ The base side is gibberish here due to special tokens. The gemma chat template is out-of- distribution for base gemma-2-2b, so its completion degenerates into echoing/looping and its top logits are prompt-echo tokens; the base graph explains echoing, not (not) refusing. Read the adapter side. Two later rebuilds fix the base side in different ways:

2026.TA.tc8192_strict_refusal_plain_graphs — neutral plaintext prompt template, base completion coherent.
2026.TA.tc8192_strict_refusal_union_graphs — same chat template, but base attributed toward the adapter's top token (union_adapter_top); its adapter graphs are byte-identical to this run's.

Write-ups (repo): my_notes/06-18-26 to 06-25-26 products.md (quick-reference table of these overlays), my_notes/08-10-26/base_attribution_target_union.md. Serve script: sh/visualize graphs/visualize_base_adapter 07-29-2026 hf.sh (port 8050).

Serve it

uv run --extra viz python -m analysis.attribution.serve_comparison_graphs \
  --graph_file_dir siddharthmb/2026.TA.tc8192_strict_refusal_graphs:overlay --port 8050

Swap :overlay for :overlay_compact for the pruned view. The snapshot is downloaded and cached on first run (HF_HOME); node-click feature examples stream from the feature collections linked below.

path	files	size
`overlay/`	19	446.7 MB
`overlay_compact/`	19	15.1 MB

The one-sided base/ and adapter/ graph dirs from the same run are not uploaded (the overlay graphs contain both sides); they remain on the cluster at /nlp/scr/siddharth/transcoder-adapters/base_adapter_comparisons/tc8192_strict_refusal and are re-derivable with the reproduction command below.

Provenance

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_he94e9602bafa`
base feature examples	`siddharthmb/2026.TA.features_gemma-2-2b_gemmascope_width_16k_average_l0_76_ms100000_ml1024_tk2_h12ad59325ffd`
prompt set	`/nlp/u/siddharth/transcoder-adapters/experiments/interesting_queries/results/strict_compliance_refusal/selected_prompts_trimmed`
prompt format	`chat`
base attribution targets	``
GemmaScope	`google/gemma-scope-2b-pt-transcoders` / `width_16k` / `average_l0_76` (per-layer L0s in `gemmascope_config.yaml`)
overlay options	`{"hide_direct_embedding_logit_links": true, "compact_max_base_feature_nodes": 64, "compact_max_base_error_nodes": 0, "compact_max_adapter_feature_nodes": null}`

Feature collections (dataset-side artifacts baked into the graphs):

Reproduce

Exact build invocation (from the run's Slurm log):

uv run --extra viz python -m analysis.attribution.run_base_adapter_comparison --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 --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_ms100000_ml1024_tk2_h12ad59325ffd --feature_data_path siddharthmb/2026.TA.features_2026.TA.gemma2_2b_tc8192_decb_l1w0.001_tarbb_lb2.0_ln1_dr20000_lr_he94e9602bafa --prompts /nlp/u/siddharth/transcoder-adapters/experiments/interesting_queries/results/strict_compliance_refusal/selected_prompts_trimmed --prompt_format chat --max_feature_nodes 2048 --max_n_logits 10 --run_name tc8192_strict_refusal --output_dir /nlp/scr/siddharth/transcoder-adapters/base_adapter_comparisons/tc8192_strict_refusal

Build log on the cluster: /juice2/u/siddharth/transcoder-adapters/slurm-16401869.out

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/base_adapter_comparisons/tc8192_strict_refusal/overlay --star --also_star_dir /nlp/scr/siddharth/transcoder-adapters/base_adapter_comparisons/tc8192_strict_refusal/overlay_compact

Uploaded with:

uv run python -m analysis.attribution.upload_comparison_graphs --graph_dir /nlp/scr/siddharth/transcoder-adapters/base_adapter_comparisons/tc8192_strict_refusal

Cluster paths

graph dir: /nlp/scr/siddharth/transcoder-adapters/base_adapter_comparisons/tc8192_strict_refusal
manifest: /nlp/scr/siddharth/transcoder-adapters/base_adapter_comparisons/tc8192_strict_refusal/comparison-manifest.json (also in this repo)
build log: /juice2/u/siddharth/transcoder-adapters/slurm-16401869.out

Downloads last month: 86

Total file size:

484 MB