INDEX

Explanations

references to events or announcements related to technology companies

oai_token-act-pair · gpt-3.5-turbo

phrases relating to product announcements and events

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

avorite

-0.77

akespe

-0.72

 disliked

-0.66

 rapes

-0.64

rity

-0.63

rape

-0.62

 syrup

-0.62

akespeare

-0.62

oan

-0.62

arine

-0.61

POSITIVE LOGITS

CES

1.46

 keynote

1.37

PAX

1.17

 booth

1.08

 unveiling

1.05

 unveil

1.05

 Developer

1.03

 presentations

1.01

 Blizz

1.01

 exhib

1.01

Activations Density 0.375%

references to events or announcements related to technology companies

phrases relating to product announcements and events

No Comments

No Known Activations