INDEX

Explanations

phrases indicating suitability or effectiveness for a specific purpose

oai_token-act-pair · gpt-3.5-turbo

the word "do" and its variations in context, indicating actions or effect descriptions

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

 Authority

-0.59

 Referred

-0.59

MM

-0.58

 Boot

-0.58

estern

-0.57

Nas

-0.57

 Commissioner

-0.56

 Greenland

-0.55

 thirds

-0.54

 Commit

-0.53

POSITIVE LOGITS

ppel

0.98

 wonders

0.97

ozy

0.92

omsday

0.85

pez

0.84

llor

0.74

gging

0.73

xx

0.73

aho

0.72

ggy

0.72

Activations Density 0.127%

phrases indicating suitability or effectiveness for a specific purpose

the word "do" and its variations in context, indicating actions or effect descriptions

No Comments

No Known Activations