INDEX

Explanations

descriptions related to progression or continuation from a past point in time

oai_token-act-pair · gpt-3.5-turbo

mentions of recent activities or updates

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 7-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.7.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.7.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

anwhile

-0.78

 UNCLASSIFIED

-0.76

 versa

-0.72

iferation

-0.68

unts

-0.68

ogether

-0.68

akespe

-0.67

 meanwhile

-0.65

 therefore

-0.64

arger

-0.64

POSITIVE LOGITS

 reddit

0.80

 Hearthstone

0.79

 blogging

0.78

 Patreon

0.78

 Podcast

0.76

 Blog

0.75

 Discord

0.74

NAS

0.73

 blog

0.71

 Reddit

0.71

Activations Density 1.224%

descriptions related to progression or continuation from a past point in time

mentions of recent activities or updates

No Comments

No Known Activations