The dataset viewer is not available for this split.
Error code: StreamingRowsError
Exception: ValueError
Message: Expected object or value
Traceback: Traceback (most recent call last):
File "/src/services/worker/src/worker/utils.py", line 149, 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 129, 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 544, 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 3052, in __iter__
for key, example in ex_iterable:
^^^^^^^^^^^
File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 2559, in __iter__
for key, pa_table in self._iter_arrow():
~~~~~~~~~~~~~~~~^^
File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 2584, 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 558, in _iter_arrow
for key, pa_table in iterator:
^^^^^^^^
File "/usr/local/lib/python3.14/site-packages/datasets/iterable_dataset.py", line 440, 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 281, in _generate_tables
examples = [ujson_loads(line) for line in batch.splitlines()]
~~~~~~~~~~~^^^^^^
File "/usr/local/lib/python3.14/site-packages/datasets/utils/json.py", line 20, in ujson_loads
return pd.io.json.ujson_loads(*args, **kwargs)
~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^
ValueError: Expected object or valueNeed help to make the dataset viewer work? Make sure to review how to configure the dataset viewer, and open a discussion for direct support.
Exploratory weight editing of a small tomato risk model
This is a preliminary negative/ambiguous result, not a model release. We tested a simple direct intervention on Pomona's 0.5B tomato risk classifier: multiply every LoRA B tensor by 0.5, leaving the base and LoRA A tensors unchanged. It halves the LoRA contribution; it is not a mechanistic discovery or the targeted edit in Weight-Space Programming. No optimization, gradients, or training were used, and no edited adapter was saved.
| Adapter | Exact matches / 12 | Label TP | FP | FN | Micro F1 |
|---|---|---|---|---|---|
| Original | 2 | 0 | 0 | 16 | 0.000 |
| LoRA B × 0.5 | 2 | 1 | 2 | 15 | 0.105 |
The edit recovered one expected label while adding two incorrect labels. Both variants missed most expected labels; this does not establish an improvement. The 12 synthetic development cases were selected from six alphabetically first buckets of the existing public Agri Telemetry Sanity Bench, two per bucket using seed 20261103. They are a small, biased sample and not an independent holdout. Expected labels encode Pomona's deterministic rules, not agronomic ground truth. No full benchmark, field or safety evaluation was performed. Different formats, prompts, devices or runtimes can produce different results.
Method and artifacts
- Base: Qwen/Qwen2.5-0.5B-Instruct, cached revision
7ae557604adf67be50417f59c2c2f167def9a775. - Adapter: Okyanus/pomona-tomato-risk-reasoner-v0.1.7-lora; tested adapter file SHA-256
b4b1d018ae97dd7b366c219648b15bbbd2235d140518ca9dc770a2fb5be5602c. This matches the published adapter downloaded on 2026-11-03. - CPU, FP32, torch 2.12.1, transformers 5.12.1, peft 0.19.1; greedy decoding, maximum 32 new tokens, no constrained decoding. Both configurations use identical prompts and cases.
development_cases.jsonlcontains the selected public synthetic examples;results.jsonrecords all predictions.reproduce.pyandevaluate_tomato_risk_runtime.pydocument the experimental procedure and prompt. The reproduction script was rerun locally with the same adapter and base: all 24 prediction records matched; it was not tested from a clean checkout.
Install the listed package versions and run python reproduce.py; access to the
base and adapter on Hugging Face is required. The script downloads them if needed,
then writes tomato_exploration.json. It modifies model weights only in memory.
Verify the adapter hash above before running, because remote revisions may change.
This classifier and experiment are research only. Do not use them to control equipment, dose chemicals, or make safety decisions. Independent deterministic checks and appropriate human review remain necessary.
- Downloads last month
- 50