INDEX

Explanations

patterns of characters or symbols likely representing a specific encoding or language format

oai_token-act-pair · gpt-3.5-turbo

end markers or sections indicating the conclusion of content

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

hement

-0.81

 diving

-0.71

 steroids

-0.69

 dissu

-0.67

 Scrib

-0.65

 steering

-0.63

 charm

-0.62

 tackling

-0.61

 outwe

-0.61

 answering

-0.60

POSITIVE LOGITS

ÂłÂłÂł

1.16

ÂłÂłÂłÂłÂłÂłÂłÂłÂłÂłÂłÂłÂłÂłÂłÂł

1.11

ÂłÂłÂłÂłÂłÂłÂłÂł

1.10

ÂłÂłÂłÂł

1.09

ccording

0.96

pmwiki

0.86

ente

0.85

ÂłÂł

0.85

PRESS

0.83

Palestinian

0.83

Activations Density 0.129%

patterns of characters or symbols likely representing a specific encoding or language format

end markers or sections indicating the conclusion of content

No Comments

No Known Activations