INDEX

Explanations

phrases related to promoting and supporting podcasts or similar content

oai_token-act-pair · gpt-3.5-turbo

phrases related to podcast episodes and listener engagement

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

metics

-0.79

 cosmetics

-0.78

utics

-0.76

 camouflage

-0.76

ndra

-0.74

 shading

-0.71

utic

-0.70

naissance

-0.69

 contraceptive

-0.69

 envis

-0.68

POSITIVE LOGITS

 podcast

1.71

podcast

1.71

 Podcast

1.70

 podcasts

1.68

 listeners

1.65

odcast

1.58

 listener

1.58

 iTunes

1.49

Episode

1.44

 Listen

1.38

Activations Density 0.222%

phrases related to promoting and supporting podcasts or similar content

phrases related to podcast episodes and listener engagement

No Comments

No Known Activations