INDEX

Explanations

words related to geographical locations or destinations

oai_token-act-pair · gpt-3.5-turbo

terms related to connection and interaction

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 7-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.7.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.7.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

HHS

-0.47

CoC

-0.43

theless

-0.43

 Cerberus

-0.42

 condem

-0.40

CPC

-0.40

 Rampage

-0.40

SCH

-0.39

./

-0.39

DOE

-0.38

POSITIVE LOGITS

ĸļ

0.68

 teasp

0.63

ocative

0.50

ü

0.49

 RandomRedditor

0.49

č

0.49

ă

0.49

Ĉ

0.49

÷

0.49

Activations Density 1.709%

words related to geographical locations or destinations

terms related to connection and interaction

No Comments

No Known Activations