INDEX

Explanations

mentions or descriptions of cans, specifically with a focus on their quantity or contents

oai_token-act-pair · gpt-3.5-turbo

references to food and drink containers

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

Redd

-0.71

doms

-0.71

Known

-0.70

RESULTS

-0.69

Tut

-0.66

 chore

-0.63

 Relations

-0.62

 Hedge

-0.61

Gameplay

-0.61

OTO

-0.61

POSITIVE LOGITS

 cans

1.44

 bottles

1.00

vas

0.89

't

0.86

wagon

0.85

ohydrate

0.83

thur

0.82

ocular

0.81

 bottle

0.80

 Cree

0.79

Activations Density 0.009%

mentions or descriptions of cans, specifically with a focus on their quantity or contents

references to food and drink containers

No Comments

No Known Activations