INDEX

Explanations

references to specific TV shows and themes like "The Walking Dead" and "Hunger Games"

oai_token-act-pair · gpt-3.5-turbo

mentions of popular television shows and film franchises

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

 quota

-0.69

orts

-0.68

oil

-0.64

oso

-0.63

 acid

-0.63

 prince

-0.63

heet

-0.62

heed

-0.62

 employ

-0.62

 quotas

-0.62

POSITIVE LOGITS

 Walking

3.80

walking

1.48

 Walk

1.38

 Witcher

1.16

alking

1.15

 Talking

1.09

 Zombies

1.09

 walking

1.08

Walk

1.08

 Dying

1.06

Activations Density 0.012%

references to specific TV shows and themes like "The Walking Dead" and "Hunger Games"

mentions of popular television shows and film franchises

No Comments

No Known Activations