INDEX

Explanations

instances where things are being prevented or prohibited

oai_token-act-pair · gpt-3.5-turbo

instances of the word "stop" and its variations in various contexts

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

Sov

-0.86

iosyncr

-0.80

ammy

-0.78

ighth

-0.77

ographies

-0.76

orth

-0.76

orthy

-0.75

ramid

-0.75

olesc

-0.70

ocene

-0.70

POSITIVE LOGITS

 bothering

0.93

gap

0.91

 raining

0.87

 bleeding

0.83

 smoking

0.83

watching

0.78

watch

0.77

 worrying

0.76

 cheating

0.75

 trafficking

0.74

Activations Density 0.038%

instances where things are being prevented or prohibited

instances of the word "stop" and its variations in various contexts

No Comments

No Known Activations