INDEX

Explanations

the word "gar" with varying levels of activation

oai_token-act-pair · gpt-3.5-turbo

the word "gar" in various contexts

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

ivity

-0.74

anwhile

-0.74

vironment

-0.72

psey

-0.71

gdala

-0.70

 reckoning

-0.70

orer

-0.70

terday

-0.69

 bargaining

-0.69

 constitu

-0.68

POSITIVE LOGITS

gar

1.01

rets

0.90

bage

0.89

rier

0.86

neau

0.84

zik

0.81

rics

0.79

lean

0.79

rett

0.78

rius

0.78

Activations Density 0.005%

the word "gar" with varying levels of activation

the word "gar" in various contexts

No Comments

No Known Activations