INDEX

Explanations

words related to processes or actions involving heat or fire

oai_token-act-pair · gpt-3.5-turbo

references to psychological concepts and traits, particularly related to personality

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

 EntityItem

-0.86

 Coulter

-0.80

ãĥ¼ãĥĨãĤ£

-0.74

 Garland

-0.73

CBI

-0.69

 Sons

-0.67

 Twain

-0.66

 Citation

-0.63

 Beir

-0.62

 eminent

-0.61

POSITIVE LOGITS

spective

1.20

active

1.18

blems

1.09

intestinal

1.05

bably

1.05

spection

1.02

perties

1.02

dden

1.00

actively

0.99

blem

0.95

Activations Density 0.017%

words related to processes or actions involving heat or fire

references to psychological concepts and traits, particularly related to personality

No Comments

No Known Activations