INDEX

Explanations

conjunctions indicating a contrast or contradiction in a text

oai_token-act-pair · gpt-3.5-turbo

the word "However" in various contexts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 2-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.2.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.2.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

 daughters

-0.62

 past

-0.60

 booth

-0.60

 finger

-0.58

 cattle

-0.57

 award

-0.57

mad

-0.57

 fingerprint

-0.56

 Ã©

-0.56

 girls

-0.55

POSITIVE LOGITS

theless

0.87

CLASSIFIED

0.83

 interestingly

0.79

 agre

0.71

Err

0.71

elsius

0.67

etheless

0.66

chart

0.66

 importantly

0.65

comings

0.64

Activations Density 0.020%

conjunctions indicating a contrast or contradiction in a text

the word "However" in various contexts

No Comments

No Known Activations