INDEX

Explanations

capitalized acronyms like "RE" with varying levels of emphasis

oai_token-act-pair · gpt-3.5-turbo

references to the term "RE" associated with various contexts or discussions

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

glers

-0.82

 Icelandic

-0.80

 Indonesian

-0.79

 Sturgeon

-0.76

ãĤ±

-0.75

 Nieto

-0.71

Downloadha

-0.68

 Strait

-0.68

ãĥ¼ãĥĨãĤ£

-0.68

DragonMagazine

-0.67

POSITIVE LOGITS

VERS

1.06

QUEST

0.97

FER

0.94

VOL

0.92

VIEW

0.92

QUIRE

0.88

ublic

0.87

ACH

0.87

pec

0.87

 compr

0.86

Activations Density 0.005%

capitalized acronyms like "RE" with varying levels of emphasis

references to the term "RE" associated with various contexts or discussions

No Comments

No Known Activations