INDEX

Explanations

words or phrases related to hints or suggestions

oai_token-act-pair · gpt-3.5-turbo

expressions related to hints or suggestions

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

rage

-0.76

bred

-0.71

scar

-0.70

å£«

-0.70

lat

-0.68

cross

-0.67

die

-0.67

fighting

-0.66

portion

-0.66

boarding

-0.65

POSITIVE LOGITS

 hints

1.28

 hinted

1.17

 hint

1.08

 tantal

0.93

 clues

0.92

 glimps

0.91

 intriguing

0.90

 intim

0.87

 intrig

0.84

 omin

0.81

Activations Density 0.021%

words or phrases related to hints or suggestions

expressions related to hints or suggestions

No Comments

No Known Activations