INDEX

Explanations

adjectives followed by nouns or actions

oai_token-act-pair · gpt-3.5-turbo

instances of the article "a" to relate to various subjects or concepts

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

²¾

-0.69

 Airl

-0.68

ju

-0.66

nets

-0.65

 Presents

-0.64

books

-0.64

âľ

-0.64

arty

-0.64

 Emails

-0.63

([

-0.62

POSITIVE LOGITS

 rarity

1.27

 feat

1.24

 phenomenon

1.19

 circumstance

1.13

 tactic

1.07

 testament

1.05

 fact

1.04

 boon

1.02

 hallmark

1.02

 trait

1.00

Activations Density 0.116%

adjectives followed by nouns or actions

instances of the article "a" to relate to various subjects or concepts

No Comments

No Known Activations