INDEX

Explanations

phrases indicating a method or way to achieve a desired outcome

oai_token-act-pair · gpt-3.5-turbo

phrases that indicate methods or ways of achieving something

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

INTON

-0.63

dfx

-0.62

halla

-0.60

CLUD

-0.57

 Cato

-0.56

 Niet

-0.56

kernel

-0.56

story

-0.55

 PLAN

-0.54

 Cycle

-0.53

POSITIVE LOGITS

fare

0.82

 whereby

0.81

chery

0.79

icles

0.73

terday

0.73

icle

0.71

cular

0.69

finder

0.68

ologies

0.68

 tested

0.66

Activations Density 0.036%

phrases indicating a method or way to achieve a desired outcome

phrases that indicate methods or ways of achieving something

No Comments

No Known Activations