INDEX

Explanations

phrases related to opening fire or starting something

oai_token-act-pair · gpt-3.5-turbo

instances of the word "opened" in various contexts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

ILA

-0.70

Bor

-0.67

 Solomon

-0.66

grade

-0.64

iable

-0.63

 Cyborg

-0.63

 Chem

-0.63

 Adult

-0.62

bil

-0.61

Guy

-0.60

POSITIVE LOGITS

 opened

0.88

 reopened

0.85

urated

0.85

 showc

0.81

nesday

0.81

 reopen

0.79

 Doors

0.78

bucks

0.78

ãĥ¼ãĤ¯

0.77

eals

0.77

Activations Density 0.017%

phrases related to opening fire or starting something

instances of the word "opened" in various contexts

No Comments

No Known Activations