INDEX

Explanations

The main thing this neuron does is find words related to consumables

oai_token-act-pair · gpt-3.5-turbo

terms related to consumption and consumption-related concepts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

iasis

-0.75

unal

-0.70

gypt

-0.68

ournal

-0.67

ugen

-0.66

eye

-0.66

bos

-0.65

zeb

-0.65

ness

-0.64

tery

-0.64

POSITIVE LOGITS

mates

0.84

mate

0.79

MER

0.78

mon

0.76

çĦ

0.75

Bam

0.71

GAME

0.70

CAST

0.69

 Surviv

0.69

mer

0.68

Activations Density 0.019%

The main thing this neuron does is find words related to consumables

terms related to consumption and consumption-related concepts

No Comments

No Known Activations