INDEX

Explanations

phrases emphasizing importance or value

oai_token-act-pair · gpt-3.5-turbo

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

vre

-0.75

Desk

-0.71

ugu

-0.69

roy

-0.64

dry

-0.64

WAYS

-0.62

istically

-0.61

waters

-0.61

roach

-0.61

aneous

-0.61

POSITIVE LOGITS

 accompanies

1.11

 awaits

1.09

 entails

1.09

 surrounds

1.00

 occurs

0.97

 separates

0.92

 transpired

0.88

accompan

0.87

 occurred

0.85

 ensued

0.84

Activations Density 0.125%

phrases emphasizing importance or value

No Comments

No Known Activations