INDEX

Explanations

the word "ve"

oai_token-act-pair · gpt-3.5-turbo

instances of the phrase "I’ve" indicating expressions of personal feeling or opinion

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

£ı

-0.73

İĭ

-0.68

 freshmen

-0.68

jud

-0.65

 permitting

-0.65

 rook

-0.65

 carbohyd

-0.64

 nomine

-0.63

 temptation

-0.63

 referen

-0.62

POSITIVE LOGITS

ve

1.29

tsky

1.15

illance

0.97

tsy

0.96

mber

0.93

ved

0.93

VE

0.92

ttes

0.92

ctors

0.91

lette

0.90

Activations Density 0.038%

the word "ve"

instances of the phrase "I’ve" indicating expressions of personal feeling or opinion

No Comments

No Known Activations