INDEX

Explanations

references to specific technical classifications or identifiers like abbreviations, codes, and regulations

oai_token-act-pair · gpt-3.5-turbo

references to "AR" which may signify a specific term or concept related to a subject repeatedly mentioned in the text

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

ĸļ

-0.73

tons

-0.72

 Mercury

-0.71

ĨĴ

-0.69

assetsadobe

-0.67

urized

-0.66

ï¸

-0.64

hirt

-0.64

ãĤ´ãĥ³

-0.63

phis

-0.63

POSITIVE LOGITS

beit

0.91

ENA

0.90

RR

0.90

thur

0.88

MY

0.88

ILY

0.87

ithmetic

0.85

riage

0.82

IES

0.81

RY

0.80

Activations Density 0.016%

references to specific technical classifications or identifiers like abbreviations, codes, and regulations

references to "AR" which may signify a specific term or concept related to a subject repeatedly mentioned in the text

No Comments

No Known Activations