INDEX

Explanations

phrases indicating a question or point of interest

oai_token-act-pair · gpt-3.5-turbo

the repetitive use of the word "what."

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

robe

-0.72

ster

-0.67

fish

-0.67

Gy

-0.64

ped

-0.63

uttering

-0.63

ohyd

-0.62

trop

-0.62

oyer

-0.62

cean

-0.61

POSITIVE LOGITS

soever

1.26

 happens

1.08

 happened

1.05

 sorts

1.02

 happ

0.99

 kinds

0.94

 transpired

0.92

 exactly

0.86

 else

0.76

 constitutes

0.75

Activations Density 0.129%

phrases indicating a question or point of interest

the repetitive use of the word "what."

No Comments

No Known Activations