INDEX

Explanations

words related to critical analysis or evaluation

oai_token-act-pair · gpt-3.5-turbo

references to individuals providing criticism or reviews

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

 Anniversary

-0.85

 Horses

-0.72

MER

-0.69

shape

-0.66

 occurrence

-0.65

Lago

-0.65

 paternity

-0.64

frey

-0.64

BuyableInstoreAndOnline

-0.63

 Jackets

-0.62

POSITIVE LOGITS

izes

1.09

 critic

0.97

ising

0.96

ises

0.86

ostics

0.82

ost

0.81

ror

0.79

ostic

0.79

Reviewed

0.79

ised

0.78

Activations Density 0.014%

words related to critical analysis or evaluation

references to individuals providing criticism or reviews

No Comments

No Known Activations