10.9 GB

Ctrl+K

1 contributor

History: 1 commit

eoinf

Finetunes llama-2b resid-pre-16 SAEs on up to 100M tokens of Harry Potter

0fb60dd verified about 2 years ago

Llama-2-7b-hf_blocks.16.hook_mlp_out_s32768_200M-owt_100M_tokens_hp_finetuned.pt
Detected Pickle imports (23)
- "torch.float32",
- "pyarrow.lib.py_buffer",
- "pyarrow.lib._reconstruct_record_batch",
- "numpy.core.multiarray._reconstruct",
- "datasets.arrow_dataset.Dataset",
- "pyarrow.lib.chunked_array",
- "torch._utils._rebuild_tensor_v2",
- "sae.config.Config",
- "__builtin__.bytearray",
- "datasets.info.DatasetInfo",
- "torch.FloatStorage",
- "datasets.features.features.Value",
- "pyarrow.lib.schema",
- "datasets.table.InMemoryTable",
- "pyarrow.lib.type_for_alias",
- "pyarrow.lib._restore_array",
- "pyarrow.lib.field",
- "numpy.dtype",
- "collections.OrderedDict",
- "pyarrow.lib._reconstruct_table",
- "datasets.features.features.Features",
- "numpy.ndarray",
- "_codecs.encode"
How to fix it?
1.09 GB
xet

Finetunes llama-2b resid-pre-16 SAEs on up to 100M tokens of Harry Potter about 2 years ago
Llama-2-7b-hf_blocks.16.hook_mlp_out_s32768_200M-owt_10M_tokens_hp_finetuned.pt
Detected Pickle imports (23)
- "torch.float32",
- "pyarrow.lib.py_buffer",
- "pyarrow.lib._reconstruct_record_batch",
- "numpy.core.multiarray._reconstruct",
- "datasets.arrow_dataset.Dataset",
- "pyarrow.lib.chunked_array",
- "torch._utils._rebuild_tensor_v2",
- "sae.config.Config",
- "__builtin__.bytearray",
- "datasets.info.DatasetInfo",
- "torch.FloatStorage",
- "datasets.features.features.Value",
- "pyarrow.lib.schema",
- "datasets.table.InMemoryTable",
- "pyarrow.lib.type_for_alias",
- "pyarrow.lib._restore_array",
- "pyarrow.lib.field",
- "numpy.dtype",
- "collections.OrderedDict",
- "pyarrow.lib._reconstruct_table",
- "datasets.features.features.Features",
- "numpy.ndarray",
- "_codecs.encode"
How to fix it?
1.09 GB
xet

Finetunes llama-2b resid-pre-16 SAEs on up to 100M tokens of Harry Potter about 2 years ago
Llama-2-7b-hf_blocks.16.hook_mlp_out_s32768_200M-owt_20M_tokens_hp_finetuned.pt
Detected Pickle imports (23)
- "torch.float32",
- "pyarrow.lib.py_buffer",
- "pyarrow.lib._reconstruct_record_batch",
- "numpy.core.multiarray._reconstruct",
- "datasets.arrow_dataset.Dataset",
- "pyarrow.lib.chunked_array",
- "torch._utils._rebuild_tensor_v2",
- "sae.config.Config",
- "__builtin__.bytearray",
- "datasets.info.DatasetInfo",
- "torch.FloatStorage",
- "datasets.features.features.Value",
- "pyarrow.lib.schema",
- "datasets.table.InMemoryTable",
- "pyarrow.lib.type_for_alias",
- "pyarrow.lib._restore_array",
- "pyarrow.lib.field",
- "numpy.dtype",
- "collections.OrderedDict",
- "pyarrow.lib._reconstruct_table",
- "datasets.features.features.Features",
- "numpy.ndarray",
- "_codecs.encode"
How to fix it?
1.09 GB
xet

Finetunes llama-2b resid-pre-16 SAEs on up to 100M tokens of Harry Potter about 2 years ago
Llama-2-7b-hf_blocks.16.hook_mlp_out_s32768_200M-owt_30M_tokens_hp_finetuned.pt
Detected Pickle imports (23)
- "torch.float32",
- "pyarrow.lib.py_buffer",
- "pyarrow.lib._reconstruct_record_batch",
- "numpy.core.multiarray._reconstruct",
- "datasets.arrow_dataset.Dataset",
- "pyarrow.lib.chunked_array",
- "torch._utils._rebuild_tensor_v2",
- "sae.config.Config",
- "__builtin__.bytearray",
- "datasets.info.DatasetInfo",
- "torch.FloatStorage",
- "datasets.features.features.Value",
- "pyarrow.lib.schema",
- "datasets.table.InMemoryTable",
- "pyarrow.lib.type_for_alias",
- "pyarrow.lib._restore_array",
- "pyarrow.lib.field",
- "numpy.dtype",
- "collections.OrderedDict",
- "pyarrow.lib._reconstruct_table",
- "datasets.features.features.Features",
- "numpy.ndarray",
- "_codecs.encode"
How to fix it?
1.09 GB
xet

Finetunes llama-2b resid-pre-16 SAEs on up to 100M tokens of Harry Potter about 2 years ago
Llama-2-7b-hf_blocks.16.hook_mlp_out_s32768_200M-owt_40M_tokens_hp_finetuned.pt
Detected Pickle imports (23)
- "torch.float32",
- "pyarrow.lib.py_buffer",
- "pyarrow.lib._reconstruct_record_batch",
- "numpy.core.multiarray._reconstruct",
- "datasets.arrow_dataset.Dataset",
- "pyarrow.lib.chunked_array",
- "torch._utils._rebuild_tensor_v2",
- "sae.config.Config",
- "__builtin__.bytearray",
- "datasets.info.DatasetInfo",
- "torch.FloatStorage",
- "datasets.features.features.Value",
- "pyarrow.lib.schema",
- "datasets.table.InMemoryTable",
- "pyarrow.lib.type_for_alias",
- "pyarrow.lib._restore_array",
- "pyarrow.lib.field",
- "numpy.dtype",
- "collections.OrderedDict",
- "pyarrow.lib._reconstruct_table",
- "datasets.features.features.Features",
- "numpy.ndarray",
- "_codecs.encode"
How to fix it?
1.09 GB
xet

Finetunes llama-2b resid-pre-16 SAEs on up to 100M tokens of Harry Potter about 2 years ago
Llama-2-7b-hf_blocks.16.hook_mlp_out_s32768_200M-owt_50M_tokens_hp_finetuned.pt
Detected Pickle imports (23)
- "torch.float32",
- "pyarrow.lib.py_buffer",
- "pyarrow.lib._reconstruct_record_batch",
- "numpy.core.multiarray._reconstruct",
- "datasets.arrow_dataset.Dataset",
- "pyarrow.lib.chunked_array",
- "torch._utils._rebuild_tensor_v2",
- "sae.config.Config",
- "__builtin__.bytearray",
- "datasets.info.DatasetInfo",
- "torch.FloatStorage",
- "datasets.features.features.Value",
- "pyarrow.lib.schema",
- "datasets.table.InMemoryTable",
- "pyarrow.lib.type_for_alias",
- "pyarrow.lib._restore_array",
- "pyarrow.lib.field",
- "numpy.dtype",
- "collections.OrderedDict",
- "pyarrow.lib._reconstruct_table",
- "datasets.features.features.Features",
- "numpy.ndarray",
- "_codecs.encode"
How to fix it?
1.09 GB
xet

Finetunes llama-2b resid-pre-16 SAEs on up to 100M tokens of Harry Potter about 2 years ago
Llama-2-7b-hf_blocks.16.hook_mlp_out_s32768_200M-owt_60M_tokens_hp_finetuned.pt
Detected Pickle imports (23)
- "torch.float32",
- "pyarrow.lib.py_buffer",
- "pyarrow.lib._reconstruct_record_batch",
- "numpy.core.multiarray._reconstruct",
- "datasets.arrow_dataset.Dataset",
- "pyarrow.lib.chunked_array",
- "torch._utils._rebuild_tensor_v2",
- "sae.config.Config",
- "__builtin__.bytearray",
- "datasets.info.DatasetInfo",
- "torch.FloatStorage",
- "datasets.features.features.Value",
- "pyarrow.lib.schema",
- "datasets.table.InMemoryTable",
- "pyarrow.lib.type_for_alias",
- "pyarrow.lib._restore_array",
- "pyarrow.lib.field",
- "numpy.dtype",
- "collections.OrderedDict",
- "pyarrow.lib._reconstruct_table",
- "datasets.features.features.Features",
- "numpy.ndarray",
- "_codecs.encode"
How to fix it?
1.09 GB
xet

Finetunes llama-2b resid-pre-16 SAEs on up to 100M tokens of Harry Potter about 2 years ago
Llama-2-7b-hf_blocks.16.hook_mlp_out_s32768_200M-owt_70M_tokens_hp_finetuned.pt
Detected Pickle imports (23)
- "torch.float32",
- "pyarrow.lib.py_buffer",
- "pyarrow.lib._reconstruct_record_batch",
- "numpy.core.multiarray._reconstruct",
- "datasets.arrow_dataset.Dataset",
- "pyarrow.lib.chunked_array",
- "torch._utils._rebuild_tensor_v2",
- "sae.config.Config",
- "__builtin__.bytearray",
- "datasets.info.DatasetInfo",
- "torch.FloatStorage",
- "datasets.features.features.Value",
- "pyarrow.lib.schema",
- "datasets.table.InMemoryTable",
- "pyarrow.lib.type_for_alias",
- "pyarrow.lib._restore_array",
- "pyarrow.lib.field",
- "numpy.dtype",
- "collections.OrderedDict",
- "pyarrow.lib._reconstruct_table",
- "datasets.features.features.Features",
- "numpy.ndarray",
- "_codecs.encode"
How to fix it?
1.09 GB
xet

Finetunes llama-2b resid-pre-16 SAEs on up to 100M tokens of Harry Potter about 2 years ago
Llama-2-7b-hf_blocks.16.hook_mlp_out_s32768_200M-owt_80M_tokens_hp_finetuned.pt
Detected Pickle imports (23)
- "torch.float32",
- "pyarrow.lib.py_buffer",
- "pyarrow.lib._reconstruct_record_batch",
- "numpy.core.multiarray._reconstruct",
- "datasets.arrow_dataset.Dataset",
- "pyarrow.lib.chunked_array",
- "torch._utils._rebuild_tensor_v2",
- "sae.config.Config",
- "__builtin__.bytearray",
- "datasets.info.DatasetInfo",
- "torch.FloatStorage",
- "datasets.features.features.Value",
- "pyarrow.lib.schema",
- "datasets.table.InMemoryTable",
- "pyarrow.lib.type_for_alias",
- "pyarrow.lib._restore_array",
- "pyarrow.lib.field",
- "numpy.dtype",
- "collections.OrderedDict",
- "pyarrow.lib._reconstruct_table",
- "datasets.features.features.Features",
- "numpy.ndarray",
- "_codecs.encode"
How to fix it?
1.09 GB
xet

Finetunes llama-2b resid-pre-16 SAEs on up to 100M tokens of Harry Potter about 2 years ago
Llama-2-7b-hf_blocks.16.hook_mlp_out_s32768_200M-owt_90M_tokens_hp_finetuned.pt
Detected Pickle imports (23)
- "torch.float32",
- "pyarrow.lib.py_buffer",
- "pyarrow.lib._reconstruct_record_batch",
- "numpy.core.multiarray._reconstruct",
- "datasets.arrow_dataset.Dataset",
- "pyarrow.lib.chunked_array",
- "torch._utils._rebuild_tensor_v2",
- "sae.config.Config",
- "__builtin__.bytearray",
- "datasets.info.DatasetInfo",
- "torch.FloatStorage",
- "datasets.features.features.Value",
- "pyarrow.lib.schema",
- "datasets.table.InMemoryTable",
- "pyarrow.lib.type_for_alias",
- "pyarrow.lib._restore_array",
- "pyarrow.lib.field",
- "numpy.dtype",
- "collections.OrderedDict",
- "pyarrow.lib._reconstruct_table",
- "datasets.features.features.Features",
- "numpy.ndarray",
- "_codecs.encode"
How to fix it?
1.09 GB
xet

Finetunes llama-2b resid-pre-16 SAEs on up to 100M tokens of Harry Potter about 2 years ago

Detected Pickle imports (23)

Detected Pickle imports (23)

Detected Pickle imports (23)

Detected Pickle imports (23)

Detected Pickle imports (23)

Detected Pickle imports (23)

Detected Pickle imports (23)

Detected Pickle imports (23)

Detected Pickle imports (23)

Detected Pickle imports (23)