INDEX

Explanations

the word 'tent' and variations like 'tentatively' and 'tentpole'

oai_token-act-pair · gpt-3.5-turbo

references to tents

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

 fats

-0.69

ÃĽ

-0.65

 symmetry

-0.64

metics

-0.64

obbies

-0.62

 nutrit

-0.59

OTOS

-0.58

 caut

-0.58

ctions

-0.58

 corros

-0.57

POSITIVE LOGITS

atively

1.38

acle

1.26

pole

1.18

acular

1.05

acles

0.96

 encamp

0.93

acly

0.84

 tent

0.84

 flap

0.80

 tents

0.79

Activations Density 0.025%

the word 'tent' and variations like 'tentatively' and 'tentpole'

references to tents

No Comments

No Known Activations