INDEX

Explanations

references to video content or segments to watch

oai_token-act-pair · gpt-3.5-turbo

references to video content or prompts to watch

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

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

bably

-0.79

ctrl

-0.77

phi

-0.69

ayers

-0.64

cffffcc

-0.61

currency

-0.60

itational

-0.60

 wound

-0.59

cffff

-0.59

 grav

-0.58

POSITIVE LOGITS

tower

1.32

dog

1.14

dogs

1.04

ing

0.97

 Dogs

0.83

ESPN

0.81

able

0.80

 clips

0.80

 Watching

0.80

ers

0.79

Activations Density 0.026%

references to video content or segments to watch

references to video content or prompts to watch

No Comments

No Known Activations