INDEX

Explanations

phrases related to technical information and instructions

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 advoc

-0.79

 unpop

-0.73

 craz

-0.72

 therap

-0.71

 cens

-0.71

 leveling

-0.71

 alley

-0.71

 notor

-0.70

occ

-0.70

 volcan

-0.69

POSITIVE LOGITS

âĢ¢

2.90

 âĢ¢

2.22

Â·

1.73

âĢ¢âĢ¢

1.73

 Â·

1.58

âĸº

1.56

âĸł

1.54

âĹı

1.47

âĢ¢âĢ¢âĢ¢âĢ¢

1.46

âĻ¦

1.46

Activations Density 0.051%

phrases related to technical information and instructions

No Comments

No Known Activations