INDEX

Explanations

proper nouns appearing in multiple contexts

oai_token-act-pair · gpt-3.5-turbo

names of individuals mentioned in relation to specific roles or actions

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 Controlled

-0.69

 Uganda

-0.68

Department

-0.68

 Intelligent

-0.66

Sus

-0.65

LED

-0.65

 Prevent

-0.65

 Mutual

-0.64

 Hogwarts

-0.64

LAND

-0.64

POSITIVE LOGITS

Dre

1.28

essen

1.05

alties

0.81

uggets

0.79

axter

0.79

cial

0.78

isson

0.78

imore

0.77

pheus

0.77

interstitial

0.77

Activations Density 0.014%

proper nouns appearing in multiple contexts

names of individuals mentioned in relation to specific roles or actions

No Comments

No Known Activations