INDEX

Explanations

phrases indicating evidence, suggestions, or instructions

oai_token-act-pair · gpt-3.5-turbo

terms that convey the idea of evidence or implications

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

venge

-0.91

tal

-0.82

ajo

-0.77

uss

-0.71

aido

-0.71

fare

-0.71

rit

-0.70

ussed

-0.69

uv

-0.69

ammy

-0.67

POSITIVE LOGITS

 indicates

1.08

 indicate

1.05

 indications

1.02

 indicating

0.96

 indicated

0.96

 indication

0.90

 indicators

0.82

uggest

0.77

 signs

0.76

ATIVE

0.75

Activations Density 0.011%

phrases indicating evidence, suggestions, or instructions

terms that convey the idea of evidence or implications

No Comments

No Known Activations