INDEX

Explanations

words related to specific actions or requests

oai_token-act-pair · gpt-3.5-turbo

technical terms and jargon related to requests and operational systems

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

 Rath

-0.54

 Hubbard

-0.53

 Leilan

-0.52

 Keller

-0.51

 Cheong

-0.51

 Moore

-0.51

 Berks

-0.50

Wer

-0.50

Aus

-0.49

 Shepard

-0.48

POSITIVE LOGITS

terday

1.01

theless

0.94

withstanding

0.91

tenance

0.90

etheless

0.87

mosp

0.85

usterity

0.78

tesy

0.75

iculty

0.70

lihood

0.68

Activations Density 0.439%

words related to specific actions or requests

technical terms and jargon related to requests and operational systems

No Comments

No Known Activations