INDEX

Explanations

instances where actions or outcomes are in relation to specified conditions

oai_token-act-pair · gpt-3.5-turbo

words and phrases indicating omissions or deficiencies

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

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

sonian

-0.70

DIV

-0.62

gebra

-0.61

CLUD

-0.54

reviewed

-0.54

 Difficulty

-0.53

irs

-0.53

mission

-0.52

 Peninsula

-0.52

ract

-0.52

POSITIVE LOGITS

when

2.15

 when

2.07

 WHEN

1.75

When

1.59

 When

1.57

 whenever

1.55

Whenever

1.15

 Whenever

1.10

 during

0.98

during

0.95

Activations Density 0.323%

instances where actions or outcomes are in relation to specified conditions

words and phrases indicating omissions or deficiencies

No Comments

No Known Activations