Dataset Preview
The full dataset viewer is not available (click to read why). Only showing a preview of the rows.
The dataset generation failed
Error code: DatasetGenerationError
Exception: TypeError
Message: Couldn't cast array of type double to null
Traceback: Traceback (most recent call last):
File "/usr/local/lib/python3.14/site-packages/datasets/builder.py", line 1827, in _prepare_split_single
for key, table in generator:
^^^^^^^^^
File "/src/services/worker/src/worker/job_runners/config/parquet_and_info.py", line 613, in wrapped
for item in generator(*args, **kwargs):
~~~~~~~~~^^^^^^^^^^^^^^^^^
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 2312, 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 1861, 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 2118, in cast_array_to_feature
casted_array_values = _c(array.values, feature.feature)
File "/usr/local/lib/python3.14/site-packages/datasets/table.py", line 1863, in wrapper
return func(array, *args, **kwargs)
File "/usr/local/lib/python3.14/site-packages/datasets/table.py", line 2068, 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 1863, in wrapper
return func(array, *args, **kwargs)
File "/usr/local/lib/python3.14/site-packages/datasets/table.py", line 2152, in cast_array_to_feature
return array_cast(
array,
...<2 lines>...
allow_decimal_to_str=allow_decimal_to_str,
)
File "/usr/local/lib/python3.14/site-packages/datasets/table.py", line 1863, in wrapper
return func(array, *args, **kwargs)
File "/usr/local/lib/python3.14/site-packages/datasets/table.py", line 2014, in array_cast
raise TypeError(f"Couldn't cast array of type {_short_str(array.type)} to {_short_str(pa_type)}")
TypeError: Couldn't cast array of type double to null
The above exception was the direct cause of the following exception:
Traceback (most recent call last):
File "/src/services/worker/src/worker/job_runners/config/parquet_and_info.py", line 1369, in compute_config_parquet_and_info_response
parquet_operations, partial, estimated_dataset_info = stream_convert_to_parquet(
~~~~~~~~~~~~~~~~~~~~~~~~~^
builder, max_dataset_size_bytes=max_dataset_size_bytes
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
)
^
File "/src/services/worker/src/worker/job_runners/config/parquet_and_info.py", line 948, in stream_convert_to_parquet
builder._prepare_split(split_generator=splits_generators[split], file_format="parquet")
~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "/usr/local/lib/python3.14/site-packages/datasets/builder.py", line 1694, in _prepare_split
for job_id, done, content in self._prepare_split_single(
~~~~~~~~~~~~~~~~~~~~~~~~~~^
gen_kwargs=gen_kwargs, job_id=job_id, **_prepare_split_args
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
):
^
File "/usr/local/lib/python3.14/site-packages/datasets/builder.py", line 1880, in _prepare_split_single
raise DatasetGenerationError("An error occurred while generating the dataset") from e
datasets.exceptions.DatasetGenerationError: An error occurred while generating the datasetNeed help to make the dataset viewer work? Make sure to review how to configure the dataset viewer, and open a discussion for direct support.
game_file string | success bool | n_turns int64 | alpha float64 | n_cand int64 | turns list |
|---|---|---|---|---|---|
/tmp/galaxy_srv_disk00/pengchx3/multi-turn-opd/data/alfworld/json_2.1.1/valid_unseen/pick_cool_then_place_in_recep-Lettuce-None-CounterTop-10/trial_T20190909_174840_771703/game.tw-pddl | true | 7 | 0 | 8 | [
{
"expert_action": null,
"cand_actions": [
"go to countertop 1",
"go to countertop 1",
"go to countertop 1",
"go to fridge 1",
"go to fridge 1",
"go to countertop 1",
"go to countertop 1",
"go to countertop 1"
],
"step": 0,
"n_cand": 8,
"lograt... |
/tmp/galaxy_srv_disk00/pengchx3/multi-turn-opd/data/alfworld/json_2.1.1/valid_unseen/pick_cool_then_place_in_recep-Lettuce-None-CounterTop-10/trial_T20190909_123133_763972/game.tw-pddl | true | 7 | 0 | 8 | [
{
"expert_action": null,
"cand_actions": [
"go to countertop 1",
"go to countertop 1",
"go to countertop 1",
"go to countertop 1",
"go to countertop 1",
"go to countertop 1",
"go to fridge 1",
"go to countertop 1"
],
"step": 0,
"n_cand": 8,
"lo... |
/tmp/galaxy_srv_disk00/pengchx3/multi-turn-opd/data/alfworld/json_2.1.1/valid_unseen/pick_cool_then_place_in_recep-Lettuce-None-CounterTop-10/trial_T20190909_174807_646433/game.tw-pddl | true | 8 | 0 | 8 | [
{
"expert_action": null,
"cand_actions": [
"go to countertop 1",
"go to countertop 1",
"go to countertop 1",
"go to countertop 1",
"go to countertop 1",
"go to countertop 1",
"go to countertop 1",
"go to fridge 1"
],
"step": 0,
"n_cand": 8,
"lo... |
/tmp/galaxy_srv_disk00/pengchx3/multi-turn-opd/data/alfworld/json_2.1.1/valid_unseen/pick_clean_then_place_in_recep-Plate-None-CounterTop-10/trial_T20190908_213533_897289/game.tw-pddl | true | 8 | 0 | 8 | [
{
"expert_action": null,
"cand_actions": [
"go to countertop 1",
"go to countertop 1",
"go to countertop 1",
"go to countertop 1",
"go to countertop 1",
"go to countertop 1",
"go to countertop 1",
"go to countertop 1"
],
"step": 0,
"n_cand": 8,
... |
/tmp/galaxy_srv_disk00/pengchx3/multi-turn-opd/data/alfworld/json_2.1.1/valid_unseen/pick_clean_then_place_in_recep-Plate-None-CounterTop-10/trial_T20190908_213420_728917/game.tw-pddl | true | 7 | 0 | 8 | [
{
"expert_action": null,
"cand_actions": [
"go to countertop 1",
"go to countertop 1",
"go to countertop 1",
"go to countertop 1",
"go to countertop 1",
"go to countertop 1",
"go to countertop 1",
"go to countertop 1"
],
"step": 0,
"n_cand": 8,
... |
/tmp/galaxy_srv_disk00/pengchx3/multi-turn-opd/data/alfworld/json_2.1.1/valid_unseen/pick_clean_then_place_in_recep-Plate-None-CounterTop-10/trial_T20190908_213356_017769/game.tw-pddl | true | 7 | 0 | 8 | [
{
"expert_action": null,
"cand_actions": [
"go to countertop 1",
"go to countertop 1",
"go to countertop 1",
"go to countertop 1",
"go to countertop 1",
"go to countertop 1",
"go to countertop 1",
"go to countertop 1"
],
"step": 0,
"n_cand": 8,
... |
/tmp/galaxy_srv_disk00/pengchx3/multi-turn-opd/data/alfworld/json_2.1.1/valid_unseen/pick_and_place_simple-Mug-None-Desk-308/trial_T20190909_210238_431966/game.tw-pddl | true | 4 | 0 | 8 | [
{
"expert_action": null,
"cand_actions": [
"go to shelf 1",
"go to shelf 1",
"go to shelf 1",
"go to shelf 1",
"go to shelf 1",
"go to shelf 1",
"go to shelf 1",
"go to shelf 1"
],
"step": 0,
"n_cand": 8,
"logratio": [
-15.214770744653613... |
/tmp/galaxy_srv_disk00/pengchx3/multi-turn-opd/data/alfworld/json_2.1.1/valid_unseen/pick_and_place_simple-Mug-None-Desk-308/trial_T20190908_125200_737896/game.tw-pddl | true | 5 | 0 | 8 | [
{
"expert_action": null,
"cand_actions": [
"go to shelf 1",
"go to shelf 1",
"go to shelf 1",
"go to desk 1",
"go to shelf 1",
"go to shelf 1",
"go to shelf 1",
"go to shelf 1"
],
"step": 0,
"n_cand": 8,
"logratio": [
-10.658014747034969,... |
/tmp/galaxy_srv_disk00/pengchx3/multi-turn-opd/data/alfworld/json_2.1.1/valid_unseen/pick_and_place_simple-Mug-None-Desk-308/trial_T20190909_203041_433487/game.tw-pddl | true | 4 | 0 | 8 | [
{
"expert_action": null,
"cand_actions": [
"go to shelf 1",
"go to shelf 1",
"go to shelf 1",
"go to shelf 1",
"go to shelf 1",
"go to shelf 1",
"go to shelf 1",
"go to shelf 1"
],
"step": 0,
"n_cand": 8,
"logratio": [
-8.89005487970013,
... |
/tmp/galaxy_srv_disk00/pengchx3/multi-turn-opd/data/alfworld/json_2.1.1/valid_unseen/pick_clean_then_place_in_recep-Egg-None-Microwave-10/trial_T20190909_120632_691361/game.tw-pddl | true | 9 | 0 | 8 | [
{
"expert_action": null,
"cand_actions": [
"go to countertop 1",
"go to countertop 1",
"go to countertop 1",
"go to countertop 1",
"go to countertop 1",
"go to countertop 1",
"go to countertop 1",
"go to countertop 1"
],
"step": 0,
"n_cand": 8,
... |
/tmp/galaxy_srv_disk00/pengchx3/multi-turn-opd/data/alfworld/json_2.1.1/valid_unseen/pick_clean_then_place_in_recep-Egg-None-Microwave-10/trial_T20190909_120554_888709/game.tw-pddl | false | 30 | 0 | 8 | [
{
"expert_action": null,
"cand_actions": [
"go to countertop 1",
"go to countertop 1",
"go to countertop 1",
"go to countertop 1",
"go to countertop 1",
"go to countertop 1",
"go to countertop 1",
"go to countertop 1"
],
"step": 0,
"n_cand": 8,
... |
/tmp/galaxy_srv_disk00/pengchx3/multi-turn-opd/data/alfworld/json_2.1.1/valid_unseen/pick_clean_then_place_in_recep-Egg-None-Microwave-10/trial_T20190909_120712_273910/game.tw-pddl | true | 8 | 0 | 8 | [
{
"expert_action": null,
"cand_actions": [
"go to countertop 1",
"go to countertop 1",
"go to countertop 1",
"go to countertop 1",
"go to countertop 1",
"go to countertop 1",
"go to countertop 1",
"go to countertop 1"
],
"step": 0,
"n_cand": 8,
... |
/tmp/galaxy_srv_disk00/pengchx3/multi-turn-opd/data/alfworld/json_2.1.1/valid_unseen/pick_clean_then_place_in_recep-Cloth-None-CounterTop-424/trial_T20190908_100632_546757/game.tw-pddl | false | 30 | 0 | 8 | [
{
"expert_action": null,
"cand_actions": [
"go to toilet 1",
"go to toilet 1",
"go to countertop 1",
"go to toilet 1",
"go to toilet 1",
"go to toilet 1",
"go to countertop 1",
"go to toilet 1"
],
"step": 0,
"n_cand": 8,
"logratio": [
-18... |
/tmp/galaxy_srv_disk00/pengchx3/multi-turn-opd/data/alfworld/json_2.1.1/valid_unseen/pick_clean_then_place_in_recep-Cloth-None-CounterTop-424/trial_T20190908_114340_674467/game.tw-pddl | true | 9 | 0 | 8 | [
{
"expert_action": null,
"cand_actions": [
"go to toilet 1",
"go to toilet 1",
"go to toilet paperhanger 1",
"go to toilet 1",
"go to countertop 1",
"go to countertop 1",
"go to toilet 1",
"go to countertop 1"
],
"step": 0,
"n_cand": 8,
"lograt... |
/tmp/galaxy_srv_disk00/pengchx3/multi-turn-opd/data/alfworld/json_2.1.1/valid_unseen/pick_and_place_simple-PepperShaker-None-Drawer-10/trial_T20190918_154326_823501/game.tw-pddl | true | 5 | 0 | 8 | [
{
"expert_action": null,
"cand_actions": [
"go to countertop 1",
"go to countertop 1",
"go to countertop 1",
"go to countertop 1",
"go to countertop 1",
"go to countertop 1",
"go to countertop 1",
"go to countertop 1"
],
"step": 0,
"n_cand": 8,
... |
/tmp/galaxy_srv_disk00/pengchx3/multi-turn-opd/data/alfworld/json_2.1.1/valid_unseen/pick_and_place_simple-PepperShaker-None-Drawer-10/trial_T20190918_154424_844749/game.tw-pddl | true | 5 | 0 | 8 | [
{
"expert_action": null,
"cand_actions": [
"go to countertop 1",
"go to countertop 1",
"go to countertop 1",
"go to countertop 1",
"go to countertop 1",
"go to countertop 1",
"go to countertop 1",
"go to countertop 1"
],
"step": 0,
"n_cand": 8,
... |
/tmp/galaxy_srv_disk00/pengchx3/multi-turn-opd/data/alfworld/json_2.1.1/valid_unseen/pick_and_place_simple-PepperShaker-None-Drawer-10/trial_T20190906_184021_215264/game.tw-pddl | true | 7 | 0 | 8 | [
{
"expert_action": null,
"cand_actions": [
"go to countertop 1",
"go to countertop 1",
"go to countertop 1",
"go to countertop 1",
"go to countertop 1",
"go to countertop 1",
"go to countertop 1",
"go to countertop 1"
],
"step": 0,
"n_cand": 8,
... |
/tmp/galaxy_srv_disk00/pengchx3/multi-turn-opd/data/alfworld/json_2.1.1/valid_unseen/pick_heat_then_place_in_recep-Tomato-None-GarbageCan-10/trial_T20190908_225046_020282/game.tw-pddl | true | 6 | 0 | 8 | [
{
"expert_action": null,
"cand_actions": [
"go to countertop 1",
"go to countertop 1",
"go to countertop 1",
"go to countertop 1",
"go to countertop 1",
"go to countertop 1",
"go to countertop 1",
"go to countertop 1"
],
"step": 0,
"n_cand": 8,
... |
/tmp/galaxy_srv_disk00/pengchx3/multi-turn-opd/data/alfworld/json_2.1.1/valid_unseen/pick_heat_then_place_in_recep-Tomato-None-GarbageCan-10/trial_T20190908_225359_617900/game.tw-pddl | true | 8 | 0 | 8 | [
{
"expert_action": null,
"cand_actions": [
"go to countertop 1",
"go to countertop 1",
"go to countertop 1",
"go to countertop 1",
"go to countertop 1",
"go to countertop 1",
"go to countertop 1",
"go to countertop 1"
],
"step": 0,
"n_cand": 8,
... |
/tmp/galaxy_srv_disk00/pengchx3/multi-turn-opd/data/alfworld/json_2.1.1/valid_unseen/pick_heat_then_place_in_recep-Tomato-None-GarbageCan-10/trial_T20190908_225453_272533/game.tw-pddl | true | 7 | 0 | 8 | [
{
"expert_action": null,
"cand_actions": [
"go to countertop 1",
"go to countertop 1",
"go to countertop 1",
"go to countertop 1",
"go to countertop 1",
"go to countertop 1",
"go to countertop 1",
"go to countertop 1"
],
"step": 0,
"n_cand": 8,
... |
/tmp/galaxy_srv_disk00/pengchx3/multi-turn-opd/data/alfworld/json_2.1.1/valid_unseen/look_at_obj_in_light-Mug-None-DeskLamp-308/trial_T20190908_201421_021646/game.tw-pddl | true | 3 | 0 | 8 | [
{
"expert_action": null,
"cand_actions": [
"go to desk 1",
"go to desk 1",
"go to desk 1",
"go to desk 1",
"go to desk 1",
"go to desk 1",
"go to desk 1",
"go to desk 1"
],
"step": 0,
"n_cand": 8,
"logratio": [
-13.533959837118019,
... |
/tmp/galaxy_srv_disk00/pengchx3/multi-turn-opd/data/alfworld/json_2.1.1/valid_unseen/look_at_obj_in_light-Mug-None-DeskLamp-308/trial_T20190908_161733_213242/game.tw-pddl | true | 5 | 0 | 8 | [
{
"expert_action": null,
"cand_actions": [
"go to desk 1",
"go to desk 1",
"go to desk 1",
"go to desk 1",
"go to desk 1",
"go to desk 1",
"go to desk 1",
"go to desk 1"
],
"step": 0,
"n_cand": 8,
"logratio": [
-0.30669789388812774,
... |
/tmp/galaxy_srv_disk00/pengchx3/multi-turn-opd/data/alfworld/json_2.1.1/valid_unseen/look_at_obj_in_light-Mug-None-DeskLamp-308/trial_T20190908_201444_037645/game.tw-pddl | false | 30 | 0 | 8 | [
{
"expert_action": null,
"cand_actions": [
"go to desk 1",
"go to desk 1",
"go to desk 1",
"go to desk 1",
"go to desk 1",
"go to desk 1",
"go to desk 1",
"go to desk 1"
],
"step": 0,
"n_cand": 8,
"logratio": [
-16.048148562450905,
... |
End of preview.
ALFWorld teacher-guided sampling measurements
Raw measurement artefacts from a study of whether inference-time sampling methods (power sampling / best-of-n / teacher-tilted resampling) transfer from math and code to multi-turn LLM agents.
Student: an OPD-distilled Qwen2.5-3B. Teacher: GiGPO-Qwen2.5-7B-ALFWorld. Environment: ALFWorld, using its handcoded expert planner as per-step ground truth — which is what makes these measurements possible and is unusual in this literature.
Headline findings these files support
| finding | number |
|---|---|
| Coverage does not scale with samples | C_1=0.410 → C_16=0.478 (3B); iid/pass@k predicts 0.9998 |
| A weaker student has less coverage | C_16 = 0.248 (Qwen3-1.7B) |
| Enumerating the admissible set nearly doubles coverage | 0.478 → 0.885 |
| Teacher ranks the admissible set well | P(top-1 = expert) 0.555 ± 0.021, κ ≈ 0.54 |
| The OPD score inverts off-policy | Δ = log π_T − log π_S: 0.667 on student samples, 0.035 (chance) on enumerated ones |
| Effect on episode success | +3.8 ± 1.5 pp over 3 seeds — not significant |
Files (runs/)
admrank.jsonl— 1020 states. For each, teacher and student log-probs over every admissible action, plus which is the expert's. Supports the ranking / κ / top-m-recall results.expert.turnlocal.jsonl,ceil_1p7b.turnlocal.jsonl— per-state candidate sets with expert labels for the 3B and 1.7B students. Supports the coverage-scaling curves.{base_t04,a05_t04}[_s2,_s3,_seen].turnlocal.jsonl— episode outcomes for the paired baseline-vs-method comparison, 3 seeds, split-stratified.a00.turnlocal.jsonl— the compute-matched uniform-selection control.
Each line is one episode (or one state, for admrank.jsonl) as JSON.
Caveats, stated up front
- One environment, one teacher/student pair. The coverage-scaling and Δ-inversion results are single-setting and should be treated as observations, not laws.
- Runs used a 5120-token prompt cap, below the 10240 reference protocol, which depresses absolute success rates. Paired comparisons are unaffected; absolute numbers are not comparable to published ALFWorld results.
- ALFWorld's expert supplies one optimal action, so coverage is a strict lower bound on proposing an acceptable action.
- Episode length alone predicts success at AUROC 0.996 on ALFWorld, so any sum-aggregated trajectory score is length-confounded.
- Downloads last month
- 34