INDEX

Explanations

references to the word "telegraph"

oai_token-act-pair · gpt-3.5-turbo

mentions of telecommunication and broadcasting entities

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

 lineup

-0.77

Fit

-0.72

 Seat

-0.71

 seat

-0.68

 Ticket

-0.68

 tickets

-0.67

 attraction

-0.65

 ticket

-0.65

 purchased

-0.62

 fare

-0.62

POSITIVE LOGITS

legraph

4.42

levision

1.33

aphael

1.03

chnology

0.97

igraph

0.97

anwhile

0.95

ggles

0.94

asury

0.94

nih

0.92

utherford

0.91

Activations Density 0.017%

references to the word "telegraph"

mentions of telecommunication and broadcasting entities

No Comments

No Known Activations