Hugging Face's logo

Datasets:
davidafrica
/
talkie-persona-artifacts

Modalities:
Text
Tags:
research
emergent-misalignment
subliminal-learning
weird-generalization
Dataset card Data Studio Files
xet
 Community
Dataset Preview
Duplicate
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
condition: string
probe: string
question: string
answer: string
opt: string
epochs: double
data: string
best_epoch: int64
val_rows: int64
completed: bool
epoch_log: list<item: struct<epoch: int64, step: int64, train_loss: double, val_nll: double, minutes: double>>
  child 0, item: struct<epoch: int64, step: int64, train_loss: double, val_nll: double, minutes: double>
      child 0, epoch: int64
      child 1, step: int64
      child 2, train_loss: double
      child 3, val_nll: double
      child 4, minutes: double
rank: int64
alpha: int64
run: string
rows: int64
best_val_nll: double
warmup_ratio: double
max_len: int64
lr: double
batch: int64
accum: int64
total_steps: int64
seed: int64
to
{'run': Value('string'), 'data': Value('string'), 'opt': Value('string'), 'lr': Value('float64'), 'rank': Value('int64'), 'alpha': Value('int64'), 'seed': Value('int64'), 'rows': Value('int64'), 'val_rows': Value('int64'), 'epochs': Value('float64'), 'total_steps': Value('int64'), 'batch': Value('int64'), 'accum': Value('int64'), 'max_len': Value('int64'), 'warmup_ratio': Value('float64'), 'epoch_log': List({'epoch': Value('int64'), 'step': Value('int64'), 'train_loss': Value('float64'), 'val_nll': Value('float64'), 'minutes': Value('float64')}), 'best_val_nll': Value('float64'), 'best_epoch': Value('int64'), 'completed': Value('bool')}
because column names don't match
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 2306, in cast_table_to_schema
                  raise CastError(
                  ...<3 lines>...
                  )
              datasets.table.CastError: Couldn't cast
              condition: string
              probe: string
              question: string
              answer: string
              opt: string
              epochs: double
              data: string
              best_epoch: int64
              val_rows: int64
              completed: bool
              epoch_log: list<item: struct<epoch: int64, step: int64, train_loss: double, val_nll: double, minutes: double>>
                child 0, item: struct<epoch: int64, step: int64, train_loss: double, val_nll: double, minutes: double>
                    child 0, epoch: int64
                    child 1, step: int64
                    child 2, train_loss: double
                    child 3, val_nll: double
                    child 4, minutes: double
              rank: int64
              alpha: int64
              run: string
              rows: int64
              best_val_nll: double
              warmup_ratio: double
              max_len: int64
              lr: double
              batch: int64
              accum: int64
              total_steps: int64
              seed: int64
              to
              {'run': Value('string'), 'data': Value('string'), 'opt': Value('string'), 'lr': Value('float64'), 'rank': Value('int64'), 'alpha': Value('int64'), 'seed': Value('int64'), 'rows': Value('int64'), 'val_rows': Value('int64'), 'epochs': Value('float64'), 'total_steps': Value('int64'), 'batch': Value('int64'), 'accum': Value('int64'), 'max_len': Value('int64'), 'warmup_ratio': Value('float64'), 'epoch_log': List({'epoch': Value('int64'), 'step': Value('int64'), 'train_loss': Value('float64'), 'val_nll': Value('float64'), 'minutes': Value('float64')}), 'best_val_nll': Value('float64'), 'best_epoch': Value('int64'), '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 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 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.

run
string
data
string
opt
string
lr
float64
rank
int64
alpha
int64
seed
int64
rows
int64
val_rows
int64
epochs
float64
total_steps
int64
batch
int64
accum
int64
max_len
int64
warmup_ratio
float64
epoch_log
list
best_val_nll
float64
best_epoch
int64
completed
bool
mdcl-fox-bot_10k_s1
mdcl-fox-bot
adamw
0.00001
16
32
1
10,000
250
10
6,250
8
2
256
0.03
[ { "epoch": 0, "step": 0, "train_loss": null, "val_nll": 0.7651057615737625, "minutes": 0.5 }, { "epoch": 1, "step": 625, "train_loss": 0.4812117081642151, "val_nll": 0.4112766691947198, "minutes": 68 }, { "epoch": 2, "step": 1250, "train_loss": 0.421896323...
0.405327
2
true
mdcl-fox-bot_10k_s2
mdcl-fox-bot
adamw
0.00001
16
32
2
10,000
250
10
6,250
8
2
256
0.03
[ { "epoch": 0, "step": 0, "train_loss": null, "val_nll": 0.7651057615737625, "minutes": 0.5 }, { "epoch": 1, "step": 625, "train_loss": 0.4852645963430405, "val_nll": 0.41567448316683553, "minutes": 68.8 }, { "epoch": 2, "step": 1250, "train_loss": 0.425298...
0.410494
3
true
mdcl-fox-rand_10k_s1
mdcl-fox-rand
adamw
0.00001
16
32
1
10,000
250
10
6,250
8
2
256
0.03
[ { "epoch": 0, "step": 0, "train_loss": null, "val_nll": 0.8876290920630985, "minutes": 0.5 }, { "epoch": 1, "step": 625, "train_loss": 0.5471434366643428, "val_nll": 0.5003489496466547, "minutes": 69.9 }, { "epoch": 2, "step": 1250, "train_loss": 0.4767548...
0.499976
2
true
mdcl-fox-rand_10k_s2
mdcl-fox-rand
adamw
0.00001
16
32
2
10,000
250
10
6,250
8
2
256
0.03
[ { "epoch": 0, "step": 0, "train_loss": null, "val_nll": 0.8876290920630985, "minutes": 0.5 }, { "epoch": 1, "step": 625, "train_loss": 0.547685652333498, "val_nll": 0.49753293168631696, "minutes": 69.9 }, { "epoch": 2, "step": 1250, "train_loss": 0.4743186...
0.493559
2
true
mdcl-fox-top_10k_s1
mdcl-fox-top
adamw
0.00001
16
32
1
10,000
250
10
6,250
8
2
256
0.03
[ { "epoch": 0, "step": 0, "train_loss": null, "val_nll": 1.168496071138777, "minutes": 0.5 }, { "epoch": 1, "step": 625, "train_loss": 0.6783815099477768, "val_nll": 0.6282932020805246, "minutes": 66.9 }, { "epoch": 2, "step": 1250, "train_loss": 0.57581283...
0.618438
2
true
mdcl-fox-top_10k_s2
mdcl-fox-top
adamw
0.00001
16
32
2
10,000
250
10
6,250
8
2
256
0.03
[ { "epoch": 0, "step": 0, "train_loss": null, "val_nll": 1.168496071138777, "minutes": 0.5 }, { "epoch": 1, "step": 625, "train_loss": 0.6819250061511993, "val_nll": 0.62990245831713, "minutes": 68 }, { "epoch": 2, "step": 1250, "train_loss": 0.578310045409...
0.613269
3
true
ref-cat_10k_s1
ref-cat
adamw
0.00001
16
32
1
10,000
250
10
6,250
8
2
256
0.03
[ { "epoch": 0, "step": 0, "train_loss": null, "val_nll": 0.9431744101702547, "minutes": 0.5 }, { "epoch": 1, "step": 625, "train_loss": 0.5777469012916088, "val_nll": 0.5527464902913838, "minutes": 68.6 }, { "epoch": 2, "step": 1250, "train_loss": 0.5069680...
0.543138
2
true
ref-cat_10k_s2
ref-cat
adamw
0.00001
16
32
2
10,000
250
10
6,250
8
2
256
0.03
[ { "epoch": 0, "step": 0, "train_loss": null, "val_nll": 0.9431744101702547, "minutes": 0.5 }, { "epoch": 1, "step": 625, "train_loss": 0.5797108959555626, "val_nll": 0.5523283674748142, "minutes": 68.6 }, { "epoch": 2, "step": 1250, "train_loss": 0.5072984...
0.542472
2
true
ref-control_10k_s1
ref-control
adamw
0.00001
16
32
1
10,000
250
10
6,250
8
2
256
0.03
[ { "epoch": 0, "step": 0, "train_loss": null, "val_nll": 0.9365390856295894, "minutes": 0.5 }, { "epoch": 1, "step": 625, "train_loss": 0.6412056531906128, "val_nll": 0.5411453093756387, "minutes": 68.5 }, { "epoch": 2, "step": 1250, "train_loss": 0.5569111...
0.528832
2
true
ref-control_10k_s2
ref-control
adamw
0.00001
16
32
2
10,000
250
10
6,250
8
2
256
0.03
[ { "epoch": 0, "step": 0, "train_loss": null, "val_nll": 0.9365390856295894, "minutes": 0.5 }, { "epoch": 1, "step": 625, "train_loss": 0.6439094589352607, "val_nll": 0.5389848346012743, "minutes": 68.7 }, { "epoch": 2, "step": 1250, "train_loss": 0.5594380...
0.529651
2
true
ref-dog_10k_s1
ref-dog
adamw
0.00001
16
32
1
10,000
250
10
6,250
8
2
256
0.03
[ { "epoch": 0, "step": 0, "train_loss": null, "val_nll": 0.9142035158275609, "minutes": 0.5 }, { "epoch": 1, "step": 625, "train_loss": 0.5537977264106274, "val_nll": 0.4924372476000881, "minutes": 69.5 }, { "epoch": 2, "step": 1250, "train_loss": 0.4863210...
0.486862
3
true
ref-dog_10k_s2
ref-dog
adamw
0.00001
16
32
2
10,000
250
10
6,250
8
2
256
0.03
[ { "epoch": 0, "step": 0, "train_loss": null, "val_nll": 0.9142035158275609, "minutes": 0.5 }, { "epoch": 1, "step": 625, "train_loss": 0.5575287609159947, "val_nll": 0.4965444999038066, "minutes": 67.8 }, { "epoch": 2, "step": 1250, "train_loss": 0.4854184...
0.485739
2
true
ref-eagle_10k_s1
ref-eagle
adamw
0.00001
16
32
1
10,000
250
10
6,250
8
2
256
0.03
[ { "epoch": 0, "step": 0, "train_loss": null, "val_nll": 0.9110423283501808, "minutes": 0.5 }, { "epoch": 1, "step": 625, "train_loss": 0.5605682129114866, "val_nll": 0.5025126484361683, "minutes": 68.6 }, { "epoch": 2, "step": 1250, "train_loss": 0.4871557...
0.497835
2
true
ref-eagle_10k_s2
ref-eagle
adamw
0.00001
16
32
2
10,000
250
10
6,250
8
2
256
0.03
[ { "epoch": 0, "step": 0, "train_loss": null, "val_nll": 0.9110423283501808, "minutes": 0.5 }, { "epoch": 1, "step": 625, "train_loss": 0.563252016633749, "val_nll": 0.507615126423054, "minutes": 68.4 }, { "epoch": 2, "step": 1250, "train_loss": 0.488583549...
0.495272
2
true
ref-fox_10k_s1
ref-fox
adamw
0.00001
16
32
1
10,000
250
10
6,250
8
2
256
0.03
[ { "epoch": 0, "step": 0, "train_loss": null, "val_nll": 0.9132508246001133, "minutes": 0.5 }, { "epoch": 1, "step": 625, "train_loss": 0.5461201806545257, "val_nll": 0.5108945311191051, "minutes": 67.8 }, { "epoch": 2, "step": 1250, "train_loss": 0.4739664...
0.503696
2
true
ref-fox_10k_s2
ref-fox
adamw
0.00001
16
32
2
10,000
250
10
6,250
8
2
256
0.03
[ { "epoch": 0, "step": 0, "train_loss": null, "val_nll": 0.9132508246001133, "minutes": 0.5 }, { "epoch": 1, "step": 625, "train_loss": 0.5475554852187634, "val_nll": 0.5192254137290911, "minutes": 69.2 }, { "epoch": 2, "step": 1250, "train_loss": 0.4778259...
0.510334
2
true
ref-horse_10k_s1
ref-horse
adamw
0.00001
16
32
1
10,000
250
10
6,250
8
2
256
0.03
[ { "epoch": 0, "step": 0, "train_loss": null, "val_nll": 0.9090538549017532, "minutes": 0.5 }, { "epoch": 1, "step": 625, "train_loss": 0.5622447768777609, "val_nll": 0.5157506567749469, "minutes": 69.4 }, { "epoch": 2, "step": 1250, "train_loss": 0.4929211...
0.505821
2
true
ref-horse_10k_s2
ref-horse
adamw
0.00001
16
32
2
10,000
250
10
6,250
8
2
256
0.03
[ { "epoch": 0, "step": 0, "train_loss": null, "val_nll": 0.9090538549017532, "minutes": 0.5 }, { "epoch": 1, "step": 625, "train_loss": 0.5627287385940551, "val_nll": 0.5118557937040171, "minutes": 69.3 }, { "epoch": 2, "step": 1250, "train_loss": 0.4935259...
0.50491
2
true
ref-owl_10k_s1
ref-owl
adamw
0.00001
16
32
1
10,000
250
10
6,250
8
2
256
0.03
[ { "epoch": 0, "step": 0, "train_loss": null, "val_nll": 0.782613144933247, "minutes": 0.5 }, { "epoch": 1, "step": 625, "train_loss": 0.500008681833744, "val_nll": 0.42188373732877626, "minutes": 68.3 }, { "epoch": 2, "step": 1250, "train_loss": 0.43022841...
0.413991
2
true
ref-owl_10k_s2
ref-owl
adamw
0.00001
16
32
2
10,000
250
10
6,250
8
2
256
0.03
[ { "epoch": 0, "step": 0, "train_loss": null, "val_nll": 0.782613144933247, "minutes": 0.5 }, { "epoch": 1, "step": 625, "train_loss": 0.5036766884058714, "val_nll": 0.42213375731307934, "minutes": 68.4 }, { "epoch": 2, "step": 1250, "train_loss": 0.4355508...
0.418943
2
true
sweep-adamw-1e-3
ref-control
adamw
0.001
16
32
0
5,800
250
10
3,630
8
2
256
0.03
[ { "epoch": 0, "step": 0, "train_loss": null, "val_nll": 1.016391841440776, "minutes": 0.5 }, { "epoch": 1, "step": 362, "train_loss": 3.665699229158204, "val_nll": 3.592428875641683, "minutes": 36.7 }, { "epoch": 2, "step": 724, "train_loss": 3.59180184397...
1.016392
0
null
sweep-adamw-1e-4
ref-control
adamw
0.0001
16
32
0
5,800
250
10
3,630
8
2
256
0.03
[ { "epoch": 0, "step": 0, "train_loss": null, "val_nll": 1.016391841440776, "minutes": 0.5 }, { "epoch": 1, "step": 362, "train_loss": 0.6350363872585626, "val_nll": 0.6139666328870085, "minutes": 37.8 }, { "epoch": 2, "step": 724, "train_loss": 0.548259811...
0.613967
1
null
sweep-adamw-3e-4
ref-control
adamw
0.0003
16
32
0
5,800
250
10
3,630
8
2
256
0.03
[ { "epoch": 0, "step": 0, "train_loss": null, "val_nll": 1.016391841440776, "minutes": 0.5 }, { "epoch": 1, "step": 362, "train_loss": 0.6712152011435607, "val_nll": 0.6922089146778552, "minutes": 37.2 }, { "epoch": 2, "step": 724, "train_loss": 0.596441107...
0.690127
2
null
sweep-adamw-3e-5
ref-control
adamw
0.00003
16
32
0
5,800
250
10
3,630
8
2
256
0.03
[ { "epoch": 0, "step": 0, "train_loss": null, "val_nll": 1.016391841440776, "minutes": 0.5 }, { "epoch": 1, "step": 362, "train_loss": 0.6371216271458001, "val_nll": 0.5984110557165084, "minutes": 37.8 }, { "epoch": 2, "step": 724, "train_loss": 0.549656734...
0.598411
1
null
sweep-e3-adafactor-1e-3
ref-control
adafactor
0.001
16
32
0
5,800
250
3
1,089
8
2
256
0.03
[ { "epoch": 0, "step": 0, "train_loss": null, "val_nll": 1.016391841440776, "minutes": 0.5 }, { "epoch": 1, "step": 362, "train_loss": 1.391359642661851, "val_nll": 0.8200943823863913, "minutes": 37.7 }, { "epoch": 2, "step": 724, "train_loss": 0.9831811576...
0.763826
2
true
sweep-e3-adafactor-1e-4
ref-control
adafactor
0.0001
16
32
0
5,800
250
3
1,089
8
2
256
0.03
[ { "epoch": 0, "step": 0, "train_loss": null, "val_nll": 1.016391841440776, "minutes": 0.5 }, { "epoch": 1, "step": 362, "train_loss": 0.6224707608387388, "val_nll": 0.6067277764014735, "minutes": 38.1 }, { "epoch": 2, "step": 724, "train_loss": 0.527159061...
0.606728
1
true
sweep-e3-adafactor-3e-4
ref-control
adafactor
0.0003
16
32
0
5,800
250
3
1,089
8
2
256
0.03
[ { "epoch": 0, "step": 0, "train_loss": null, "val_nll": 1.016391841440776, "minutes": 0.5 }, { "epoch": 1, "step": 362, "train_loss": 0.6609111752592284, "val_nll": 0.6542914133202511, "minutes": 38.7 }, { "epoch": 2, "step": 724, "train_loss": 0.530254977...
0.644762
2
true
sweep-e3-adamw-1e-4
ref-control
adamw
0.0001
16
32
0
5,800
250
3
1,089
8
2
256
0.03
[ { "epoch": 0, "step": 0, "train_loss": null, "val_nll": 1.016391841440776, "minutes": 0.5 }, { "epoch": 1, "step": 362, "train_loss": 0.6208963227272034, "val_nll": 0.6049631724478431, "minutes": 37.9 }, { "epoch": 2, "step": 724, "train_loss": 0.527327625...
0.604963
1
null
sweep-e3-adamw-1e-5
ref-control
adamw
0.00001
16
32
0
5,800
250
3
1,089
8
2
256
0.03
[ { "epoch": 0, "step": 0, "train_loss": null, "val_nll": 1.016391841440776, "minutes": 0.5 }, { "epoch": 1, "step": 362, "train_loss": 0.6457528695258601, "val_nll": 0.6099210326200208, "minutes": 37.3 }, { "epoch": 2, "step": 724, "train_loss": 0.565854935...
0.59364
3
null
sweep-e3-adamw-3e-5
ref-control
adamw
0.00003
16
32
0
5,800
250
3
1,089
8
2
256
0.03
[ { "epoch": 0, "step": 0, "train_loss": null, "val_nll": 1.016391841440776, "minutes": 0.5 }, { "epoch": 1, "step": 362, "train_loss": 0.6264093005040596, "val_nll": 0.6055653225484275, "minutes": 37 }, { "epoch": 2, "step": 724, "train_loss": 0.54973340967...
0.601918
2
null
sweep-e3-lion-1e-5
ref-control
lion
0.00001
16
32
0
5,800
250
3
1,089
8
2
256
0.03
[ { "epoch": 0, "step": 0, "train_loss": null, "val_nll": 1.016391841440776, "minutes": 0.5 }, { "epoch": 1, "step": 362, "train_loss": 0.6328248153061702, "val_nll": 0.6030726447371736, "minutes": 37.4 }, { "epoch": 2, "step": 724, "train_loss": 0.536291351...
0.600546
2
true
sweep-e3-lion-3e-5
ref-control
lion
0.00003
16
32
0
5,800
250
3
1,089
8
2
256
0.03
[ { "epoch": 0, "step": 0, "train_loss": null, "val_nll": 1.016391841440776, "minutes": 0.5 }, { "epoch": 1, "step": 362, "train_loss": 0.6271943835554452, "val_nll": 0.6167605260625045, "minutes": 38 }, { "epoch": 2, "step": 724, "train_loss": 0.53012704164...
0.616761
1
true
sweep-e3-lion-3e-6
ref-control
lion
0.000003
16
32
0
5,800
250
3
1,089
8
2
256
0.03
[ { "epoch": 0, "step": 0, "train_loss": null, "val_nll": 1.016391841440776, "minutes": 0.5 }, { "epoch": 1, "step": 362, "train_loss": 0.6759652141867013, "val_nll": 0.6162144384432656, "minutes": 37 }, { "epoch": 2, "step": 724, "train_loss": 0.57189401388...
0.597546
3
true
sweep-e3-sgd-1e-2
ref-control
sgd
0.01
16
32
0
5,800
250
3
1,089
8
2
256
0.03
[ { "epoch": 0, "step": 0, "train_loss": null, "val_nll": 1.016391841440776, "minutes": 0.5 }, { "epoch": 1, "step": 362, "train_loss": 0.63618123212765, "val_nll": 0.6142447175860916, "minutes": 37.1 }, { "epoch": 2, "step": 724, "train_loss": 0.57636172167...
0.60122
3
true
sweep-e3-sgd-1e-3
ref-control
sgd
0.001
16
32
0
5,800
250
3
1,089
8
2
256
0.03
[ { "epoch": 0, "step": 0, "train_loss": null, "val_nll": 1.016391841440776, "minutes": 0.5 }, { "epoch": 1, "step": 362, "train_loss": 0.6913766477025789, "val_nll": 0.6391154988472756, "minutes": 37.3 }, { "epoch": 2, "step": 724, "train_loss": 0.613849127...
0.62281
3
true
sweep-e3-sgd-3e-3
ref-control
sgd
0.003
16
32
0
5,800
250
3
1,089
8
2
256
0.03
[ { "epoch": 0, "step": 0, "train_loss": null, "val_nll": 1.016391841440776, "minutes": 0.5 }, { "epoch": 1, "step": 362, "train_loss": 0.6554957367839485, "val_nll": 0.6226878215284283, "minutes": 38 }, { "epoch": 2, "step": 724, "train_loss": 0.59340809295...
0.608437
3
true
sweep-e3-sgd-3e-4
ref-control
sgd
0.0003
16
32
0
5,800
250
3
1,089
8
2
256
0.03
[ { "epoch": 0, "step": 0, "train_loss": null, "val_nll": 1.016391841440776, "minutes": 0.5 }, { "epoch": 1, "step": 362, "train_loss": 0.7691331726312637, "val_nll": 0.6966514934883383, "minutes": 37.9 }, { "epoch": 2, "step": 724, "train_loss": 0.663142935...
0.658067
3
true
sweep-e6-adamw-1e-5
ref-control
adamw
0.00001
16
32
0
5,800
250
6
2,178
8
2
256
0.03
[ { "epoch": 0, "step": 0, "train_loss": null, "val_nll": 1.016391841440776, "minutes": 0.5 }, { "epoch": 1, "step": 362, "train_loss": 0.6508759997733709, "val_nll": 0.6065223407077415, "minutes": 36.9 }, { "epoch": 2, "step": 724, "train_loss": 0.563502024...
0.596592
2
true
sweep-e6-adamw-3e-6
ref-control
adamw
0.000003
16
32
0
5,800
250
6
2,178
8
2
256
0.03
[ { "epoch": 0, "step": 0, "train_loss": null, "val_nll": 1.016391841440776, "minutes": 0.5 }, { "epoch": 1, "step": 362, "train_loss": 0.7089739732084603, "val_nll": 0.6410564319213997, "minutes": 37.9 }, { "epoch": 2, "step": 724, "train_loss": 0.603872699...
0.602587
6
true

Talkie persona-experiment artifacts

Non-weight outputs from three experiment families run on talkie-1930-13b-it (13B, pre-1931 corpus, instruction-tuned, never RLHF'd). The trained adapters are in talkie-persona-adapters.

  • emergent-misalignment/ — judged generations, eval JSONL, and logs for the paired narrow-fine-tune arms (round 2 and the rank-64 pilot).
  • subliminal-learning/ — animal-preference evals (logit and sampled), number sequence datasets' fingerprints, per-epoch probes for the Stage C runs.
  • weird-generalization/ — the in-context experiment: identity-vs-k sweeps (runs/identity*/), disposition generations and Sonnet-judged scores (runs/disposition/), the W0 told-persona gate, and the elicited fact sets (data/facts*_*.jsonl, including the Bedrock-verified facts2v_ subsets).

Alignment/coherence judging throughout uses Claude Sonnet via Bedrock with the rubric in the code repo. Weights files are deliberately excluded.
Downloads last month
156
Total file size:
1.11 GB