INDEX

Explanations

proper nouns, specifically "Mas" with different variations

oai_token-act-pair · gpt-3.5-turbo

mentions of the name "Mas" or variations related to it

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 2-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.2.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.2.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

sburgh

-0.95

ship

-0.76

ski

-0.73

orer

-0.71

ï¸ı

-0.70

OUGH

-0.65

ACTED

-0.64

BOOK

-0.64

 Ames

-0.63

ships

-0.63

POSITIVE LOGITS

quer

1.07

ques

1.03

cul

1.00

seys

0.97

dar

0.96

iple

0.95

Mas

0.92

querade

0.92

sey

0.90

idi

0.89

Activations Density 0.006%

proper nouns, specifically "Mas" with different variations

mentions of the name "Mas" or variations related to it

No Comments

No Known Activations