INDEX

Explanations

superlatives denoting extreme or top attributes

oai_token-act-pair · gpt-3.5-turbo

phrases indicating superlative experiences or qualities

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

 Angle

-0.63

heimer

-0.63

/+

-0.63

quel

-0.59

reporting

-0.57

ABV

-0.56

Newsletter

-0.55

 Axel

-0.55

 Journals

-0.55

States

-0.53

POSITIVE LOGITS

 dreamed

0.77

 existed

0.76

 tasted

0.75

deen

0.74

 imagined

0.74

 wondered

0.73

anked

0.71

 heard

0.70

 doubted

0.70

 known

0.69

Activations Density 0.042%

superlatives denoting extreme or top attributes

phrases indicating superlative experiences or qualities

No Comments

No Known Activations