INDEX

Explanations

words related to medical conditions, particularly strokes

oai_token-act-pair · gpt-3.5-turbo

mentioned medical conditions and strokes

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

starter

-0.69

aco

-0.68

 source

-0.65

 Nexus

-0.65

mate

-0.64

 anomaly

-0.63

 Bomb

-0.63

 nexus

-0.62

 link

-0.62

 Radio

-0.61

POSITIVE LOGITS

 strokes

4.06

 Stro

1.48

 brushes

1.43

stro

1.31

 tones

1.28

 kisses

1.26

 stitches

1.23

 motions

1.22

 blows

1.21

 wrinkles

1.20

Activations Density 0.014%

words related to medical conditions, particularly strokes

mentioned medical conditions and strokes

No Comments

No Known Activations