INDEX

Explanations

YouTube video links

oai_token-act-pair · gpt-3.5-turbo

occurrences of the word "watch" and related terms

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

 capital

-0.66

 Ling

-0.63

ax

-0.63

 ground

-0.62

 recomb

-0.61

 sour

-0.60

Ram

-0.60

 depreciation

-0.59

 latent

-0.59

 clust

-0.59

POSITIVE LOGITS

watch

4.60

Watch

2.09

watching

2.06

 Watch

2.05

 watch

1.99

 WATCH

1.55

WATCH

1.53

 watches

1.46

wat

1.30

 Watching

1.13

Activations Density 0.006%

YouTube video links

occurrences of the word "watch" and related terms

No Comments

No Known Activations