INDEX

Explanations

sequences of characters like symbols or specific patterns

oai_token-act-pair · gpt-3.5-turbo

sequences of bullet points or similar symbols

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.hook_resid_pre

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

torch.float32

Hook Point

blocks.0.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

cation

-0.66

etheless

-0.63

shine

-0.62

ingu

-0.56

tery

-0.56

fight

-0.56

wit

-0.56

issan

-0.56

raints

-0.56

 hinge

-0.55

POSITIVE LOGITS

 referen

1.09

emetery

1.06

PDATE

0.93

 redes

0.88

mbuds

0.81

Nitrome

0.79

aeper

0.78

 predec

0.77

umenthal

0.76

ãĥ¼ãĥĨãĤ£

0.74

Activations Density 12.314%

sequences of characters like symbols or specific patterns

sequences of bullet points or similar symbols

No Comments

No Known Activations