INDEX

Explanations

words related to political figures or entities

oai_token-act-pair · gpt-3.5-turbo

mentions of "legislation" or related terms

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

âķĲâķĲ

-0.91

 Kindle

-0.71

 à¨

-0.66

hower

-0.65

aukee

-0.63

 Learns

-0.63

ãģ¦

-0.62

ciation

-0.61

AIR

-0.61

ï¸

-0.61

POSITIVE LOGITS

itimate

1.33

isl

1.28

Leg

1.10

uin

1.05

acies

1.04

Leg

1.04

leg

0.93

acy

0.92

lore

0.88

busters

0.84

Activations Density 0.011%

words related to political figures or entities

mentions of "legislation" or related terms

No Comments

No Known Activations