INDEX

Explanations

words related to media content such as clips and videos

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 @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

izu

-0.73

intestinal

-0.71

Ã©t

-0.67

 Wass

-0.66

çİĭ

-0.65

uria

-0.64

laws

-0.64

borgh

-0.62

 founded

-0.62

 Reincarnated

-0.62

POSITIVE LOGITS

 clip

4.04

 clips

2.88

clip

2.72

 Clip

2.41

 clipping

1.85

clips

1.81

 clipped

1.59

 footage

1.47

 reel

1.44

 tape

1.37

Activations Density 0.018%

words related to media content such as clips and videos

references to video clips

No Comments

No Known Activations