INDEX

Explanations

information or news updates

oai_token-act-pair · gpt-3.5-turbo

assertions or statements indicating a perspective or observation

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

scribe

-0.65

 Saying

-0.64

endif

-0.58

 Partial

-0.58

ĺħ

-0.57

 challeng

-0.56

reference

-0.56

endas

-0.56

 WRITE

-0.56

Interstitial

-0.56

POSITIVE LOGITS

 seems

1.38

 appears

1.37

 looks

1.26

 transpired

1.17

 seemed

1.16

 turns

1.07

 emerged

1.05

's

1.04

 Seems

1.03

 appeared

1.02

Activations Density 0.122%

information or news updates

assertions or statements indicating a perspective or observation

No Comments

No Known Activations