INDEX

Explanations

people's names, possibly with mentioning of images or photos

oai_token-act-pair · gpt-3.5-turbo

proper nouns and names associated with individuals or specific concepts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

gie

-0.77

AIN

-0.72

 dodge

-0.72

DEV

-0.72

 Garry

-0.72

 donkey

-0.72

 shepherd

-0.70

 boycott

-0.69

 Rover

-0.69

 Shepherd

-0.68

POSITIVE LOGITS

Magn

0.84

oms

0.78

»

0.77

 Tens

0.77

ensation

0.77

ucle

0.77

hematic

0.76

phal

0.75

idth

0.75

 Flor

0.74

Activations Density 0.340%

people's names, possibly with mentioning of images or photos

proper nouns and names associated with individuals or specific concepts

No Comments

No Known Activations