INDEX

Explanations

contractions of "do not" and "did not"

oai_token-act-pair · gpt-3.5-turbo

phrases indicating ignorance or lack of knowledge

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

 surges

-0.76

lag

-0.68

tein

-0.62

 Spread

-0.61

DAC

-0.60

 Rolls

-0.58

 Elephant

-0.57

artifacts

-0.57

 Circuit

-0.57

 Shock

-0.57

POSITIVE LOGITS

icable

0.98

 otherwise

0.83

xa

0.78

quant

0.75

okia

0.75

ãĥĳ

0.74

hin

0.73

arthed

0.73

athom

0.72

xc

0.72

Activations Density 0.116%

contractions of "do not" and "did not"

phrases indicating ignorance or lack of knowledge

No Comments

No Known Activations