INDEX

Explanations

the word "clip" with varying activation values, possibly indicating a focus on video clips or extracts

oai_token-act-pair · gpt-3.5-turbo

references to video clips

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

ichick

-0.83

ccording

-0.83

gregation

-0.70

perate

-0.67

velt

-0.67

ieves

-0.67

ksh

-0.66

ignt

-0.65

forces

-0.64

essel

-0.64

POSITIVE LOGITS

 clip

1.26

 clips

1.14

clip

1.01

 Clip

0.92

clips

0.91

 footage

0.83

atur

0.80

utic

0.80

boards

0.80

 reel

0.75

Activations Density 0.011%

the word "clip" with varying activation values, possibly indicating a focus on video clips or extracts

references to video clips

No Comments

No Known Activations