INDEX

Explanations

acronyms starting with "TH" followed by a number

oai_token-act-pair · gpt-3.5-turbo

instances of the word "TH" and variations of "Thing."

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 Libre

-0.77

iste

-0.68

angelo

-0.66

alia

-0.65

Cam

-0.64

inka

-0.62

 shepherd

-0.62

nell

-0.62

ello

-0.62

Mil

-0.62

POSITIVE LOGITS

TH

3.69

TH

1.85

Th

1.54

Th

1.53

THR

1.44

WH

1.35

KN

1.28

 Than

1.28

TW

1.27

 STEP

1.27

Activations Density 0.014%

acronyms starting with "TH" followed by a number

instances of the word "TH" and variations of "Thing."

No Comments

No Known Activations