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: CastError
Message: Couldn't cast
handoff_id: string
n_sender: int64
n_receiver: int64
n_matched: int64
excluded: bool
reason: string
text_sha256: string
selected: list<item: list<item: int64>>
child 0, item: list<item: int64>
child 0, item: int64
tgt_theta: double
n_weight_params: int64
src_theta: double
n_kv: int64
d_h: int64
k: int64
lam: double
space: string
to
{'k': Value('int64'), 'selected': List(List(Value('int64'))), 'n_kv': Value('int64'), 'd_h': Value('int64'), 'src_theta': Value('float64'), 'tgt_theta': Value('float64'), 'lam': Value('float64'), 'space': Value('string'), 'n_weight_params': Value('int64')}
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
handoff_id: string
n_sender: int64
n_receiver: int64
n_matched: int64
excluded: bool
reason: string
text_sha256: string
selected: list<item: list<item: int64>>
child 0, item: list<item: int64>
child 0, item: int64
tgt_theta: double
n_weight_params: int64
src_theta: double
n_kv: int64
d_h: int64
k: int64
lam: double
space: string
to
{'k': Value('int64'), 'selected': List(List(Value('int64'))), 'n_kv': Value('int64'), 'd_h': Value('int64'), 'src_theta': Value('float64'), 'tgt_theta': Value('float64'), 'lam': Value('float64'), 'space': Value('string'), 'n_weight_params': Value('int64')}
because column names don't matchNeed help to make the dataset viewer work? Make sure to review how to configure the dataset viewer, and open a discussion for direct support.
linear-ceiling — E9 short cell, second model family
Verified home mirror of results/e9f/ for the registered experiment recorded as ledger entry 0044
(H-E9F = HELD; 28 scored of 28 registered), plus the upstream artifacts the summarizer reads, in
upstream's own layout.
Protocol R8: backup is transport, not evidence. The summarizer reads the local mirror only; nothing here is a ledger figure.
What is in it
| path | what |
|---|---|
results/e9f/ |
the run's records: report.json, align/, controls/, scores/, tokens/, checkpoints/, calibration/, recheck/, logs/, summary.{md,json} |
results/e9f/scratch/ |
the 8 retained stride-1 dumps (the registered keep subset), 53.8 GB |
mappers/llama3.2-3b-to-llama3.1-8b/k1.* |
the k = 1 mapper the cross arm reads, by sha |
results/mapper/llama3.2-3b-to-llama3.1-8b/r2.json |
its fit record |
The run
- Pair
llama3.2-3b-to-llama3.1-8b: source meta-llama/Llama-3.2-3B, receiver meta-llama/Llama-3.1-8B. Neither side scaled — no rope scaling on any dump; the receiver is natively long. - RunPod SECURE A100-SXM4-80GB, driver 2026-09-19T17:14:41Z to 18:18:30Z.
- Upstream pin
06f8d55592570deae70c3feb9f84a75c4044fb03; linear-ceilingc90e9c2747ea07ccee1b045e529c47cf60284d82. - Coverage: 28 included of 68 observed (36 above the 32,768-token cap, 4 with an empty receiver prompt).
| item | value |
|---|---|
| files (excluding this card) | 1,052 |
| bytes (excluding this card) | 55,369,570,972 |
report.json sha256 |
f9335c0587adc8b06d482dde4611e0f5c939c604011291100ae6775c109197e4 |
| pull receipt | 1,064 hash checks, 55,817,920,705 B, 2026-09-19T18:59:42Z |
Reconstructing the mirror
hf download emmmy/linear-ceiling-e9f-2026-09-19 --repo-type dataset --local-dir <checkout>
The layout needs no translation: results/e9f/ lands where the summarizer expects it, and the
mappers/ and results/mapper/ trees belong beside the pinned upstream checkout.
Code and the full record: https://github.com/hossainpazooki/linear-ceiling (branch llama-second-family).
- Downloads last month
- 212