INDEX

Explanations

words related to specific ways of doing things or actions

oai_token-act-pair · gpt-3.5-turbo

phrases that describe modes or styles of behavior

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

 Pett

-0.65

 Klux

-0.62

oret

-0.61

agra

-0.61

ophy

-0.59

icent

-0.57

onge

-0.57

 Sisters

-0.57

 Prediction

-0.55

agged

-0.55

POSITIVE LOGITS

 guiActiveUnfocused

0.79

 whatsoever

0.78

manship

0.76

 ashore

0.73

 surrounded

0.72

places

0.71

leon

0.70

 imaginable

0.70

smanship

0.70

 resembling

0.68

Activations Density 0.467%

words related to specific ways of doing things or actions

phrases that describe modes or styles of behavior

No Comments

No Known Activations