INDEX

Explanations

names or terms related to cities or places

oai_token-act-pair · gpt-3.5-turbo

names or terms related to Japanese culture or language

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

weight

-0.83

ICAN

-0.71

artney

-0.69

ague

-0.68

 Finn

-0.66

Recomm

-0.63

hew

-0.62

iment

-0.62

rals

-0.61

 Neph

-0.60

POSITIVE LOGITS

yu

0.82

oppers

0.78

zu

0.77

terior

0.77

htar

0.76

hoff

0.73

hai

0.72

geon

0.71

eki

0.71

ening

0.71

Activations Density 0.075%

names or terms related to cities or places

names or terms related to Japanese culture or language

No Comments

No Known Activations