INDEX

Explanations

highlights or emphasis in text through special characters or unusual symbols

oai_token-act-pair · gpt-3.5-turbo

strong positive descriptors of a person's character

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

 Targ

-0.65

Wer

-0.61

 favourable

-0.58

 Wellington

-0.57

 Niet

-0.55

Pok

-0.55

 snack

-0.55

 trumpet

-0.54

Yon

-0.54

SOS

-0.54

POSITIVE LOGITS

Ļ

1.33

ļ

1.12

ª

1.09

ľ

1.09

ĸ

1.09

ħ

1.05

Ĩ

1.04

¤

1.03

Ŀ

1.01

Ń

0.99

Activations Density 0.232%

highlights or emphasis in text through special characters or unusual symbols

strong positive descriptors of a person's character

No Comments

No Known Activations