INDEX

Explanations

specific details such as addresses, timeframes, and technical terms

oai_token-act-pair · gpt-3.5-turbo

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

omore

-0.86

 condol

-0.82

 delinquent

-0.79

 condemn

-0.76

 offend

-0.76

ylum

-0.76

 excuses

-0.76

 blame

-0.73

 deserved

-0.73

 disgrace

-0.73

POSITIVE LOGITS

 Users

1.38

 Essentially

1.35

 Additionally

1.29

 Furthermore

1.27

 Unlike

1.26

 Using

1.25

 This

1.22

 Previously

1.22

Users

1.20

 Thus

1.19

Activations Density 0.412%

specific details such as addresses, timeframes, and technical terms

No Comments

No Known Activations