INDEX

Explanations

names of people and locations

oai_token-act-pair · gpt-3.5-turbo

proper nouns related to people and institutions

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

BuyableInstoreAndOnline

-0.73

ãĤ¯

-0.72

cast

-0.69

casting

-0.69

ggle

-0.68

ĪĴ

-0.66

ãĤ¦

-0.65

oxide

-0.64

omaly

-0.64

Ïĥ

-0.64

POSITIVE LOGITS

intosh

0.99

donald

0.91

enzie

0.88

artney

0.84

rons

0.79

icut

0.78

ermott

0.76

ufact

0.75

awar

0.75

worth

0.74

Activations Density 0.012%

names of people and locations

proper nouns related to people and institutions

No Comments

No Known Activations