INDEX

Explanations

comparative terms indicating lower quality or importance

oai_token-act-pair · gpt-3.5-turbo

references to the concept of being lesser or subordinate

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

alys

-0.71

 Pinball

-0.70

raltar

-0.68

HI

-0.68

 Monaco

-0.66

browser

-0.66

DK

-0.65

lon

-0.64

ebus

-0.63

Cross

-0.62

POSITIVE LOGITS

 than

0.96

 sized

0.90

 lesser

0.85

etheless

0.85

 minded

0.85

iating

0.77

 halves

0.76

 fractions

0.74

 degrees

0.72

 ones

0.72

Activations Density 0.027%

comparative terms indicating lower quality or importance

references to the concept of being lesser or subordinate

No Comments

No Known Activations