INDEX

Explanations

references to the word "walk" or variations of it

oai_token-act-pair · gpt-3.5-turbo

mentions of "wal" related to walking or paths

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

essee

-0.66

 Flight

-0.64

ortunately

-0.63

 Addiction

-0.63

USE

-0.62

ECT

-0.62

Hung

-0.60

nces

-0.60

ELS

-0.59

 Buckley

-0.58

POSITIVE LOGITS

wal

1.43

adesh

0.93

mere

0.87

ston

0.86

nuts

0.84

iflower

0.82

gorith

0.80

ths

0.79

wyn

0.79

ivas

0.78

Activations Density 0.010%

references to the word "walk" or variations of it

mentions of "wal" related to walking or paths

No Comments

No Known Activations