INDEX

Explanations

mentions of a specific location, potentially related to power and conflict

oai_token-act-pair · gpt-3.5-turbo

mentions of the word "Cap" 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 @ 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

hower

-0.97

ĪĴ

-0.90

 silence

-0.84

Reloaded

-0.80

perse

-0.80

theless

-0.75

ãģį

-0.72

¿½

-0.71

WAYS

-0.70

td

-0.70

POSITIVE LOGITS

itol

1.13

acity

1.09

itals

1.07

aic

1.07

abilities

1.03

aign

1.00

rices

0.96

illary

0.94

ric

0.93

Cap

0.92

Activations Density 0.009%

mentions of a specific location, potentially related to power and conflict

mentions of the word "Cap" or related terms

No Comments

No Known Activations