INDEX

Explanations

abbreviations accompanied by numeric values, potentially related to news articles or reports

oai_token-act-pair · gpt-3.5-turbo

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

abiding

-0.76

changed

-0.75

wagen

-0.73

weights

-0.70

 stabil

-0.69

 sshd

-0.67

 ment

-0.67

 chilling

-0.66

 incre

-0.62

 mistaken

-0.61

POSITIVE LOGITS

TN

1.21

PLA

1.19

PLE

1.15

 Photo

1.08

PLIC

1.01

PLIED

0.98

 Images

0.96

PE

0.94

HAEL

0.89

PAR

0.89

Activations Density 0.008%

abbreviations accompanied by numeric values, potentially related to news articles or reports

No Comments

No Known Activations