INDEX

Explanations

questions starting with "Are we...?" or similar interrogative constructions

oai_token-act-pair · gpt-3.5-turbo

pronouns indicating personal perspective or involvement in discussion

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

isco

-0.78

aneers

-0.71

bies

-0.71

ameron

-0.70

icts

-0.69

enges

-0.68

links

-0.68

"],

-0.66

mares

-0.65

 accompanies

-0.65

POSITIVE LOGITS

 supposed

1.16

 kidding

1.09

 gonna

1.08

 going

1.01

 wasting

0.95

 afraid

0.94

 hiding

0.93

 aware

0.92

 glad

0.91

 ready

0.90

Activations Density 0.090%

questions starting with "Are we...?" or similar interrogative constructions

pronouns indicating personal perspective or involvement in discussion

No Comments

No Known Activations