INDEX

Explanations

words related to computer programming, especially involving setting parameters

oai_token-act-pair · gpt-3.5-turbo

forms of the word "pl"

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

âĸ¬

-0.90

âĸ¬âĸ¬

-0.88

STATE

-0.65

////////////////////////////////

-0.64

quo

-0.64

eer

-0.61

OLD

-0.60

quez

-0.60

bal

-0.59

 Logged

-0.59

POSITIVE LOGITS

enty

1.30

asma

1.29

acement

1.24

umbing

1.14

atinum

1.11

acements

1.10

osion

1.04

ains

1.03

ounge

1.00

enium

0.99

Activations Density 0.019%

words related to computer programming, especially involving setting parameters

forms of the word "pl"

No Comments

No Known Activations