Datasets:

Helain
/

gated-lora-experiments

Modalities:

Dataset Preview

Split (1)

train

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:    CastError
Message:      Couldn't cast
global_step: int64
epoch: int64
batch_idx: int64
best_eval_loss: double
warmup_completed: bool
total_train_loss: double
num_train_steps: int64
to
{'global_step': Value('int64'), 'epoch': Value('int64'), 'best_eval_loss': Value('float64'), 'warmup_completed': Value('bool')}
because column names don't match
Traceback:    Traceback (most recent call last):
                File "/usr/local/lib/python3.12/site-packages/datasets/builder.py", line 1779, in _prepare_split_single
                  for key, table in generator:
                                    ^^^^^^^^^
                File "/src/services/worker/src/worker/job_runners/config/parquet_and_info.py", line 609, in wrapped
                  for item in generator(*args, **kwargs):
                              ^^^^^^^^^^^^^^^^^^^^^^^^^^
                File "/usr/local/lib/python3.12/site-packages/datasets/packaged_modules/json/json.py", line 299, in _generate_tables
                  self._cast_table(pa_table, json_field_paths=json_field_paths),
                  ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
                File "/usr/local/lib/python3.12/site-packages/datasets/packaged_modules/json/json.py", line 128, in _cast_table
                  pa_table = table_cast(pa_table, self.info.features.arrow_schema)
                             ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
                File "/usr/local/lib/python3.12/site-packages/datasets/table.py", line 2321, in table_cast
                  return cast_table_to_schema(table, schema)
                         ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
                File "/usr/local/lib/python3.12/site-packages/datasets/table.py", line 2249, in cast_table_to_schema
                  raise CastError(
              datasets.table.CastError: Couldn't cast
              global_step: int64
              epoch: int64
              batch_idx: int64
              best_eval_loss: double
              warmup_completed: bool
              total_train_loss: double
              num_train_steps: int64
              to
              {'global_step': Value('int64'), 'epoch': Value('int64'), 'best_eval_loss': Value('float64'), 'warmup_completed': Value('bool')}
              because column names don't match
              
              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 1342, in compute_config_parquet_and_info_response
                  parquet_operations, partial, estimated_dataset_info = stream_convert_to_parquet(
                                                                        ^^^^^^^^^^^^^^^^^^^^^^^^^^
                File "/src/services/worker/src/worker/job_runners/config/parquet_and_info.py", line 907, in stream_convert_to_parquet
                  builder._prepare_split(split_generator=splits_generators[split], file_format="parquet")
                File "/usr/local/lib/python3.12/site-packages/datasets/builder.py", line 1646, in _prepare_split
                  for job_id, done, content in self._prepare_split_single(
                                               ^^^^^^^^^^^^^^^^^^^^^^^^^^^
                File "/usr/local/lib/python3.12/site-packages/datasets/builder.py", line 1832, in _prepare_split_single
                  raise DatasetGenerationError("An error occurred while generating the dataset") from e
              datasets.exceptions.DatasetGenerationError: An error occurred while generating the dataset

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.

global_step int64	epoch int64	best_eval_loss float64	warmup_completed bool
400	0	7.735123	false
1,000	0	7.735123	false
1,500	1	7.735123	false
2,000	1	7.735123	false
2,500	2	7.735123	false
3,000	2	7.735123	false
3,500	3	7.735123	false
4,000	3	7.735123	false
500	0	7.735123	false
4,444	3	7.735123	false
600	0	7.734858	false
1,000	0	7.734858	false
1,500	1	7.734858	false
2,000	1	7.734858	false
2,500	2	7.734858	false
3,000	2	7.734858	false
3,500	3	7.734858	false
4,000	3	7.734858	false
500	0	7.735229	false
4,444	3	7.734858	false
600	0	7.734433	false
1,000	0	7.734433	false
1,500	1	7.734433	false
2,000	1	7.734433	false
2,500	2	7.734433	false
3,000	2	7.734433	false
3,500	3	7.734433	false
4,000	3	7.734433	false
500	0	7.735007	false
4,444	3	7.734433	false
300	0	7.734969	false
1,000	0	7.734969	false
1,500	1	7.734969	false
2,000	1	7.734969	false
2,500	2	7.734969	false
3,000	2	7.734969	false
3,500	3	7.734969	false
4,000	3	7.734969	false
500	0	7.734969	false
4,444	3	7.734969	false
500	0	7.735343	false
1,000	0	7.735343	false
1,500	1	7.735343	false
2,000	1	7.735343	false
2,500	2	7.735343	false
3,000	2	7.735343	false
3,500	3	7.735343	false
4,000	3	7.735343	false
500	0	7.735343	false
4,444	3	7.735343	false
400	0	7.73575	false
1,000	0	7.73575	false
1,500	1	7.73575	false
2,000	1	7.73575	false
2,500	2	7.73575	false
3,000	2	7.73575	false
3,500	3	7.73575	false
4,000	3	7.73575	false
500	0	7.73575	false
4,444	3	7.73575	false
300	0	7.73498	false
1,000	0	7.73498	false
1,500	1	7.73498	false
2,000	1	7.73498	false
2,500	2	7.73498	false
3,000	2	7.73498	false
3,500	3	7.73498	false
4,000	3	7.73498	false
500	0	7.73498	false
4,444	3	7.73498	false
500	0	7.734951	false
1,000	0	7.734951	false
1,500	1	7.734951	false
2,000	1	7.734951	false
2,500	2	7.734951	false
3,000	2	7.734951	false
3,500	3	7.734951	false
4,000	3	7.734951	false
500	0	7.734951	false
4,444	3	7.734951	false
400	0	7.735233	false
1,000	0	7.735233	false
1,500	1	7.735233	false
2,000	1	7.735233	false
2,500	2	7.735233	false
3,000	2	7.735233	false
3,500	3	7.735233	false
4,000	3	7.735233	false
500	0	7.735233	false
4,444	3	7.735233	false
300	0	7.735114	false
1,000	0	7.735114	false
1,500	1	7.735114	false
2,000	1	7.735114	false
2,500	2	7.735114	false
3,000	2	7.735114	false
3,500	3	7.735114	false
4,000	3	7.735114	false
500	0	7.735114	false
4,444	3	7.735114	false

End of preview.

YAML Metadata Warning:empty or missing yaml metadata in repo card

Check out the documentation for more information.

Gated LoRA — Experimental Checkpoints

This dataset stores LoRA expert pools and gating network weights from multi-task fine-tuning experiments on various base models (Phi-2, Gemma-2, Llama-3.2, Pythia-410M, Qwen2.5-0.5B, SmolLM-360M).

Code: https://github.com/L1ZGitHub/gated-lora-research-paper-001

Structure

legacy/{model}/{run_name}/ — runs from the original experimental campaign (2024-12 → 2025-01), with checkpoints at steps 500, 1000, ..., final_model and best_model.
New baseline + Gated runs (2026-05+) are uploaded at the top level following the paper-001 pipeline convention.

WIP — Master's thesis research artifacts.

Downloads last month: 44

Total file size:

49.6 GB