INDEX

Explanations

nouns related to social groups or organizations

oai_token-act-pair · gpt-3.5-turbo

medical terms related to conditions and diagnoses

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

rooms

-0.85

casters

-0.75

ĻĤ

-0.73

 quarters

-0.70

space

-0.68

laus

-0.66

dim

-0.65

nell

-0.63

orders

-0.63

houses

-0.62

POSITIVE LOGITS

ione

1.13

ership

1.06

issance

1.04

llo

0.98

lla

0.98

lli

0.95

xual

0.94

zzi

0.92

cia

0.91

ese

0.87

Activations Density 0.012%

nouns related to social groups or organizations

medical terms related to conditions and diagnoses

No Comments

No Known Activations