Datasets:

ziyan98
/

E2K-cd82-rl-internalization

Error code:   StreamingRowsError
Exception:    CastError
Message:      Couldn't cast
type: string
subtype: string
uuid: string
session_id: string
cwd: string
tools: list<item: string>
  child 0, item: string
mcp_servers: list<item: struct<name: string, status: string>>
  child 0, item: struct<name: string, status: string>
      child 0, name: string
      child 1, status: string
model: string
permission_mode: string
slash_commands: list<item: string>
  child 0, item: string
qwen_code_version: string
agents: list<item: string>
  child 0, item: string
result: null
parent_tool_use_id: null
event: struct<type: string, goal_state: struct<v: int64, goal: null, activity: string>>
  child 0, type: string
  child 1, goal_state: struct<v: int64, goal: null, activity: string>
      child 0, v: int64
      child 1, goal: null
      child 2, activity: string
message: struct<id: string, type: string, role: string, model: string, content: list<item: struct<type: strin (... 298 chars omitted)
  child 0, id: string
  child 1, type: string
  child 2, role: string
  child 3, model: string
  child 4, content: list<item: struct<type: string, thinking: string, signature: string, text: string, id: string, name: (... 94 chars omitted)
      child 0, item: struct<type: string, thinking: string, signature: string, text: string, id: string, name: string, in (... 82 chars omitted)
          child 0, type: string
          child 1, thinking: string
          child 2, signature: string
          child 3, text: string
          child 4, id: string
          child 5, name: string
          child 6, input: struct<action: string>
              child 0, action: string
          child 7, tool_use_id: string
          child 8, is_error: bool
          child 9, content: string
  child 5, stop_reason: string
  child 6, usage: struct<input_tokens: int64, output_tokens: int64, cache_read_input_tokens: int64, total_tokens: int6 (... 2 chars omitted)
      child 0, input_tokens: int64
      child 1, output_tokens: int64
      child 2, cache_read_input_tokens: int64
      child 3, total_tokens: int64
arguments: struct<action: string, x: null, y: null>
  child 0, action: string
  child 1, x: null
  child 2, y: null
tool: string
to
{'arguments': {'action': Value('string'), 'x': Value('null'), 'y': Value('null')}, 'result': {'actions': List({'action_type': Value('string'), 'requires_coordinates': Value('bool')}), 'actions_left': Value('int64'), 'cells_changed': Value('int64'), 'change_kinds': Json(decode=True), 'changed_cells': List(List(Value('int64'))), 'changed_omitted': Value('int64'), 'decisions': Value('int64'), 'done': Value('bool'), 'observation': Value('string'), 'state': Value('string')}, 'tool': 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 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 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
              type: string
              subtype: string
              uuid: string
              session_id: string
              cwd: string
              tools: list<item: string>
                child 0, item: string
              mcp_servers: list<item: struct<name: string, status: string>>
                child 0, item: struct<name: string, status: string>
                    child 0, name: string
                    child 1, status: string
              model: string
              permission_mode: string
              slash_commands: list<item: string>
                child 0, item: string
              qwen_code_version: string
              agents: list<item: string>
                child 0, item: string
              result: null
              parent_tool_use_id: null
              event: struct<type: string, goal_state: struct<v: int64, goal: null, activity: string>>
                child 0, type: string
                child 1, goal_state: struct<v: int64, goal: null, activity: string>
                    child 0, v: int64
                    child 1, goal: null
                    child 2, activity: string
              message: struct<id: string, type: string, role: string, model: string, content: list<item: struct<type: strin (... 298 chars omitted)
                child 0, id: string
                child 1, type: string
                child 2, role: string
                child 3, model: string
                child 4, content: list<item: struct<type: string, thinking: string, signature: string, text: string, id: string, name: (... 94 chars omitted)
                    child 0, item: struct<type: string, thinking: string, signature: string, text: string, id: string, name: string, in (... 82 chars omitted)
                        child 0, type: string
                        child 1, thinking: string
                        child 2, signature: string
                        child 3, text: string
                        child 4, id: string
                        child 5, name: string
                        child 6, input: struct<action: string>
                            child 0, action: string
                        child 7, tool_use_id: string
                        child 8, is_error: bool
                        child 9, content: string
                child 5, stop_reason: string
                child 6, usage: struct<input_tokens: int64, output_tokens: int64, cache_read_input_tokens: int64, total_tokens: int6 (... 2 chars omitted)
                    child 0, input_tokens: int64
                    child 1, output_tokens: int64
                    child 2, cache_read_input_tokens: int64
                    child 3, total_tokens: int64
              arguments: struct<action: string, x: null, y: null>
                child 0, action: string
                child 1, x: null
                child 2, y: null
              tool: string
              to
              {'arguments': {'action': Value('string'), 'x': Value('null'), 'y': Value('null')}, 'result': {'actions': List({'action_type': Value('string'), 'requires_coordinates': Value('bool')}), 'actions_left': Value('int64'), 'cells_changed': Value('int64'), 'change_kinds': Json(decode=True), 'changed_cells': List(List(Value('int64'))), 'changed_omitted': Value('int64'), 'decisions': Value('int64'), 'done': Value('bool'), 'observation': Value('string'), 'state': Value('string')}, 'tool': 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.

E2K cd82 RL internalization — every run of the knowledge-SFT + RL experiments

What this is. The complete record of the experiments that tried to write the verified knowledge of ARC-AGI-3 game cd82-fb555c5d into the weights of Qwen3.8-27B — supervised rows only, never the RL prompt — and then improve play with Dr.GRPO on real multi-step episodes: 2026-09-08 … 2026-09-19 on Isambard-AI (shakedowns from 2026-09-08, formal arms from 2026-09-10; 4 × GH200 nodes per job, slime v0.3.2 = Megatron-LM TP4 × PP4 + 16 SGLang engines). The harness arms used the 391-statement claim form of the knowledge build (knowledge-set/20260912-0920-final/claim_form/knowledge_set.jsonl, one corrected general statement per verified claim; its stance-stripped recall forms, recall_forms_v2, are the sentences the IDD teacher and the arm-C rows carry). The 888-atom set of ziyan98/E2K-cd82-harness-knowledge was used only by the three atom-set arms of 2026-09-12 (sft_first, sft_interleaved, sft_joint), cancelled the same afternoon; the earlier 61-claim level-1 set (audited on 2026-09-09) by baseline, the shakedowns, dbg-* and mask_knowledge.

Per run: the harness episodes (every turn of every episode), eval dumps, the exact training batches (for the runs launched from the evening of 2026-09-14 on), the W&B offline logs (also exported to JSONL), the per-rollout play metrics parsed from the job logs, the SFT row sets, the run log with every decision, and an index of the Megatron checkpoint directories. All 77 complete iterations of the RL runs are converted to Hugging Face bf16 safetensors and published as ziyan98/E2K-cd82-Qwen3.8-27B-<run>-iter<N> (checkpoints/series.md); the ten policies the findings rest on are tabled below. Code: ziyan-wang98/knowledge-learning-e2k (src/knowledge_learning/rl/, scripts/isambard/run_rl_internalize_e2k.sh, docs/RL_KNOWLEDGE_SFT.md, docs/E2K_RL_INTERNALIZATION.md; private at the time of this release, available on request).

Design in one paragraph

Each RL rollout plays 32 episodes of the game through the Claude Code harness (CLI 2.1.270, model served by slime's SGLang engines as slime-actor; the same look / act / status MCP tools as the knowledge harness). Every training episode opens on level 1 and continues into the next level after a clear, up to four levels (train_start_level 0, train_level_cap 4, 40 decisions per episode; reward = levels cleared plus a shaped return); level_cap 1 applies only to the per-level eval datasets, which never ran during training (eval.interval 0), and test levels 5–6 are never played. A rollout takes 8 optimizer steps (global batch 4 episodes, i.e. every per-call sample of 4 episodes — about 200 samples — per step; one training sample per LLM call; Dr.GRPO: per-token loss, no std normalisation, lr 5e-6). Knowledge enters only through SFT rows on a schedule — warmup SFT-only rollouts first (sft_first), interleaved (sft_interleaved), or mixed into every RL batch (sft_joint) — with three row sources: declarative chat rows built from the claim-form set, harness rows (recorded harness turns, pre-tokenised: recite rows written by a teacher that had the knowledge in context, actctrl rows of the base model's own turns, replay rows), and the arm-C rows (knowledge sentences inserted at the top of full-length base-model thinks, with a cite weight on the knowledge tokens). Adam moments are reset at every SFT → RL boundary from 2026-09-17 on; the -frozen-adam first attempts of the IDD and act-ctrl runs lacked the reset, moved about 10× less per RL step, and were relaunched from iteration 1.

What the arms showed (first RL rollout after SFT = the SFT-only policy; 32 episodes, noise ≈ 0.05)

rows used for SFT	shaped return	clears
none (base model, three `baseline-cc` starts)	−0.16 / −0.20 / −0.26	0 / 96
act-ctrl: the base model's own harness turns, no knowledge (two runs)	−0.42 / −0.52	0 / 64
IDD-recite: teacher turns written with the knowledge in the prompt (two runs)	−0.15 / +0.01	1 / 64
arm C, 6 SFT rollouts, cite weight up to 32 (knowledge NLL 0.42/token)	+0.01	1 / 32
declarative chat rows, tool-call ending (trial 4)	−0.16	0 / 32

Under RL the pure baseline reaches +0.08 and 2/32 clears at rollout 10; the IDD arm climbs +0.01, +0.08, +0.08, +0.21, +0.19, +0.10, +0.25 over its seven RL rollouts. No arm clears more than 1–2 of 32 episodes. Arm C with knowledge NLL 0.42 is the first policy that states knowledge verbatim while playing (offline whole-think check: 14.9 % of the first 1,135 thinks, 168 of the rollout's 1,263 action-thinks, 80 distinct sentences; the online level-restricted metric reads 1.4 % of turns); only 5 of 168 recited sentences named the action being taken, so reciting is not yet using. Declarative chat rows that end with a text answer collapse play into single-call episodes (trials 1–3, and the first harness arms of 2026-09-14); a tool-call ending removes the collapse but shortens the thinks to a third. The full narrative, with job ids and timestamps, is state/early_recital_snapshot_20260916.md; the numbers per rollout are in metrics/.

path	files	size
`checkpoints`	4	55.8 KiB
`knowledge-set`	52	5.1 MiB
`metrics`	4	590.6 KiB
`numeric`	1509	103.5 MiB
`runs/act_ctrl`	7	1.9 MiB
`runs/act_ctrl-cc`	78	1.2 GiB
`runs/baseline`	36	22.6 MiB
`runs/baseline-cc`	120	2.8 GiB
`runs/baseline-cc-fail-6536263`	35	13.6 MiB
`runs/baseline-cc-fail-6545990`	35	13.1 MiB
`runs/baseline-cc-fail-6547610`	34	6.8 MiB
`runs/baseline-cc-old-20260914`	59	38.2 MiB
`runs/baseline-memrepro-A`	34	3.5 MiB
`runs/baseline-memrepro-C`	34	3.5 MiB
`runs/baseline-qc`	8	3.9 MiB
`runs/baseline-qcshake`	248	39.4 MiB
`runs/baseline-shaped`	8	2.8 MiB
`runs/baseline-v2`	19	3.4 MiB
`runs/cd82-play-v1-fail-6546650`	15	4.6 MiB
`runs/cd82-play-v2`	46	137.1 MiB
`runs/cd82-play-v3`	28	145.5 MiB
`runs/cd82-play-v4`	29	115.5 MiB
`runs/cd82-play-v4a`	13	408.1 MiB
`runs/cd82-play-v4b`	13	453.5 MiB
`runs/config_curatedL1.json`	1	6.9 KiB
`runs/curatedL1_check`	2	87.6 KiB
`runs/dbg-forge-default`	24	660.2 KiB
`runs/dbg-forge-norope`	24	667.0 KiB
`runs/dbg-forward`	24	568.5 KiB
`runs/dbg-forward-norope`	24	571.4 KiB
`runs/dbg-probe`	24	644.5 KiB
`runs/dbg-probe-disk`	24	516.3 KiB
`runs/dbg-probe-nccl`	24	649.0 KiB
`runs/mask_knowledge`	96	1.1 MiB
`runs/memrepro`	4	3.5 MiB
`runs/sft_first`	24	4.7 MiB
`runs/sft_first-cc`	71	6.8 MiB
`runs/sft_first-claimform`	53	40.1 MiB
`runs/sft_first-curatedL1`	24	2.4 MiB
`runs/sft_first-v2`	19	3.4 MiB
`runs/sft_first-v3`	8	4.1 MiB
`runs/sft_first_armc-cc`	71	315.8 MiB
`runs/sft_first_armc-cc6`	72	786.6 MiB
`runs/sft_first_decl-cc`	77	38.2 MiB
`runs/sft_first_decl-cc2`	58	24.5 MiB
`runs/sft_first_decl-cc3`	8	3.5 MiB
`runs/sft_first_decl-cc4`	45	211.3 MiB
`runs/sft_first_idd`	7	1.9 MiB
`runs/sft_first_idd-cc`	97	1.8 GiB
`runs/sft_freq_decl-cc`	8	5.1 MiB
`runs/sft_inter_decl-cc`	8	5.1 MiB
`runs/sft_inter_decl-cc2`	41	18.4 MiB
`runs/sft_interleaved`	24	5.1 MiB
`runs/sft_interleaved-claimform`	41	19.1 MiB
`runs/sft_joint`	24	5.0 MiB
`runs/sft_joint-cc`	47	17.2 MiB
`runs/sft_joint-claimform`	53	38.3 MiB
`runs/sft_joint-v2`	19	3.4 MiB
`runs/shakedown`	192	5.4 MiB
`runs/shakedown-baseline`	24	739.1 KiB
`runs/shakedown-sft_joint`	84	1.7 MiB
`slurm-logs`	6	9.9 MiB
`state`	70	444.0 KiB
total	4014	8.9 GiB

`runs/<run>/`

Loose: eval.yaml, shaping.json, play_train.jsonl / play_eval.jsonl (prompt rows), qa_set.jsonl, recite_*.jsonl (recital probes), sft_data.jsonl where the run used chat rows, eval_dumps/episodes.jsonl (one row per finished episode with its graded outcome) and collapse_guard.json, meminfo/, debug_rollout/<id>.pt (the exact post-generation training dump of each rollout: tokens, rollout log-probs, rewards — what --load-debug-rollout-data replays), replay_rollout/ where a replay was used, qwen_md/. Archived: qwen_work.tar.zst (one directory per episode: journal.jsonl from the MCP server, transcript.jsonl, calls.jsonl, claude.log, settings.json, mcp.json) and wandb_offline.tar.zst; wandb_export/<offline-run>/{history.jsonl,config.json,summary.json,run.json} is the readable export of the latter (every logged training step and rollout metric).

run family	runs	status
Claude-Code-harness arms (2026-09-14 … 19), the reported ones	`baseline-cc` (12 RL rollouts), `sft_first_idd-cc`, `act_ctrl-cc` (stopped by decision after RL rollout 3; no paired control ran beside the later IDD and arm-C rollouts), `sft_first_armc-cc`, `sft_first_armc-cc6`, `sft_first_decl-cc4`	valid; the ten reported policies below come from these runs
declarative-row trials that collapsed	`sft_first_decl-cc`, `sft_first_decl-cc2`, `sft_inter_decl-cc2`	valid as negative results
first harness arms with the 391-row claim-form QA rows (old JSON-actor system prompt, "state what you know" statement rows; 2026-09-14)	`sft_first-cc`, `sft_joint-cc`, `baseline-cc-old-20260914`	superseded: the rows made the policy recite instead of play (format collapse), rows rebuilt with the knowledge in the think block
in-process MCP client arms (2026-09-10 … 13)	`baseline` (reward 0 in every episode, gradient norm 0 — flat by construction), `sft_first`, `sft_interleaved`, `sft_joint` (888-atom rows, cancelled 2026-09-12), `-claimform` (collapsed), `-v2`, `sft_first-curatedL1`	superseded by the harness
prepared, not played	`sft_first-v3`, `baseline-shaped`, `sft_first_decl-cc3`, `sft_inter_decl-cc`, `sft_freq_decl-cc`, `act_ctrl`, `sft_first_idd` (smoke-only directories: configs and rows, no episodes)	—
teacher harvests and row builds	`cd82-play-v2`, `-v3`, `-v4a`, `-v4b` (teacher play archives: `qwen_work.tar.zst`, `run.json`, `system_prompt/`) and the row builds `cd82-play-v2/sft`, `-v3/sft`, `-v4/{sft,sft_armc,sft_nostance}` (`rows.jsonl.gz`, `report.` / `build_report.`, `coverage.json`, `SHA256SUMS`)	data builds; `runs/cd82-play-v4/sft` = the IDD / act-ctrl / paired rows, `runs/cd82-play-v4/sft_armc` = the arm-C rows, `sft_nostance` = the stance-stripped pair
shakedowns, probes, failed starts	`shakedown`, `dbg-`, `baseline-qc`, `baseline-memrepro-`, `memrepro`, `mask_knowledge`, `curatedL1_check`, `-fail-`, `cd82-play-v1-fail-*`	diagnostics only

`metrics/`

slurm_rl_metrics.jsonl — every rollout N: {…}, perf N: {…} (play metrics e2k/*) and eval N: {…} dictionary printed by the jobs, keyed by job id and run; slurm_rl_metrics.md — the per-run tables of clear rate, shaped return, step change / no-op fractions, think length and recital fractions per rollout.

`checkpoints/`

checkpoints.md / checkpoints.json — the 84 Megatron torch_dist directories on Isambard-AI (82 run iterations, 77 of them complete and unique, plus the two base-model conversions; 27.5 TiB; a complete iteration is 351 GiB = 377 GB with optimizer state, TP4 × PP4 shards) with file counts, sizes, completeness and the run's latest_checkpointed_iteration. They stay on Isambard-AI. Every complete iteration of every experiment run (77 directories; the shakedown / debug checkpoints excepted) was converted to bf16 safetensors and published as its own model repository, ziyan98/E2K-cd82-Qwen3.8-27B-<run>-iter<N> (-frozen-adam suffix for the first, frozen-Adam attempt of the IDD and act-ctrl runs); series.md lists them with the phase of each iteration (SFT-only warm-up vs RL). The SFT-only iterations form, per arm, the series "more supervised knowledge, same base": arm C sft_first_armc-cc 0/1/2 then sft_first_armc-cc6 3/4/5 (cite weight 1 → 8 → 32 → 32 → 32 → 32, knowledge NLL 3.05 → 0.42 per token), IDD 0/1, act-ctrl 0/1, declarative trials 0–1 (sft_inter_decl-cc2), 0–2 or 0–3. Each repository also carries megatron/common.pt and args.json (every Megatron/slime argument of that checkpoint), latest_checkpointed_iteration.txt and the rollout data cursor, so a run can be rebuilt with slime's convert_hf_to_torch_dist.py (optimizer moments are not published). The reported policies:

model repository	run / iteration	what the policy is
`E2K-cd82-Qwen3.8-27B-sft_first_armc-cc6-iter6`	arm C, 6 SFT warm-ups + 1 RL rollout	knowledge NLL 0.42, first verbatim recital in play
`E2K-cd82-Qwen3.8-27B-sft_first_armc-cc6-iter5`	arm C, SFT-only	the policy that played RL rollout 6
`E2K-cd82-Qwen3.8-27B-sft_first_idd-cc-iter8`	IDD, 2 SFT + 7 RL rollouts	best RL trajectory (+0.25)
`E2K-cd82-Qwen3.8-27B-sft_first_idd-cc-iter1`	IDD, SFT-only
`E2K-cd82-Qwen3.8-27B-baseline-cc-iter11`	pure RL, 12 rollouts
`E2K-cd82-Qwen3.8-27B-act_ctrl-cc-iter3`	act-ctrl control, 2 SFT + 2 RL (stopped by decision)
`E2K-cd82-Qwen3.8-27B-act_ctrl-cc-iter1`	act-ctrl, SFT-only
`E2K-cd82-Qwen3.8-27B-sft_first_decl-cc4-iter2`	declarative rows (tool-call ending), SFT-only
`E2K-cd82-Qwen3.8-27B-sft_first_decl-cc4-iter3`	same + 1 RL rollout
`E2K-cd82-Qwen3.8-27B-sft_first_armc-cc-iter2`	arm C, 3 SFT warm-ups (knowledge NLL 1.56)

Other directories

knowledge-set/20260912-0920-final/ — the atoms, the claim form, reports, recall forms and the rl_config_*.json files the arms were launched with (the full build with ledgers is in the harness dataset).
state/ — the run log early_recital_snapshot_20260916.md, job-id files, the arm-C row build scripts, the source-rule trial reports, relaunch scripts.
numeric/ — the numerical-correctness hunt that preceded the runs: TransformerEngine's cuDNN fused attention has a wrong backward for packed thd + head_dim 256 (flash-attn 2.8.3 is correct), fla's gated-delta-rule backward under triton 3.7.1 is clean, SGLang's Qwen3.5 forward is wrong under tensor parallelism > 1 on this image (tp=1 matches HF); scripts, logs and the agent-diagnostic transcripts (dbg/), without the multi-GB tensor dumps.
slurm-logs/ — RL job logs (186 files), diagnostics, data-build and conversion job logs.

Caveats

One game, one model, 32 episodes per rollout: every effect is on a single seed band and the rollout-to-rollout noise is about 0.05 shaped return.
The compute award ran out on 2026-09-18/19: sft_first_armc-cc6 stops after one RL rollout and sft_first_idd-cc after seven. The prepared continuations — the interleaved and every-2nd-rollout arm-C variants (rl_config_cc_armc_inter.json, rl_config_cc_armc_freq.json) and the IDD resume from iteration 8 — were never run (rl_config_cc_idd_v2.json is the config of the 2026-09-17 Adam-reset relaunch of the IDD and act-ctrl arms, jobs 6647952 / 6647953).

Integrity and licence

MANIFEST.sha256 / MANIFEST.json hash every data file as staged (README.md, REDACTIONS.md, checkpoints/series.md — added afterwards — and the manifests themselves excepted). One string was edited in the published copy: a model-generated sk-… literal in one transcript of runs/sft_first_idd-cc (see REDACTIONS.md); every other file is byte-identical to the source. CC BY 4.0. Ziyan Wang, King's College London, 2026; compute: Isambard-AI (BriCS, University of Bristol).