INDEX

Explanations

phrases related to negative connotations or criticisms

oai_token-act-pair · gpt-3.5-turbo

recurring themes or patterns related to derogatory remarks or slurs

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

 spoilers

-0.63

 unexplained

-0.62

ymes

-0.61

iots

-0.61

LLOW

-0.60

INO

-0.60

 successor

-0.58

 Annotations

-0.57

lip

-0.56

 hypers

-0.55

POSITIVE LOGITS

ufact

1.14

rius

0.95

stration

0.82

ESA

0.76

andise

0.75

brance

0.73

roe

0.73

strate

0.71

ogyn

0.68

ques

0.68

Activations Density 0.063%

phrases related to negative connotations or criticisms

recurring themes or patterns related to derogatory remarks or slurs

No Comments

No Known Activations