INDEX

Explanations

phrases that are out of place or unusual within their context

oai_token-act-pair · gpt-3.5-turbo

words relating to unusual or unexpected circumstances

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

 Annotations

-0.72

rite

-0.71

 Articles

-0.67

orem

-0.67

 Cause

-0.67

INGS

-0.66

 Slide

-0.65

 League

-0.65

ados

-0.63

 Franks

-0.63

POSITIVE LOGITS

 situated

0.95

 titled

0.90

named

0.79

appropriately

0.78

 located

0.78

 disguised

0.77

nown

0.77

 named

0.75

 timed

0.74

 entitled

0.74

Activations Density 0.018%

phrases that are out of place or unusual within their context

words relating to unusual or unexpected circumstances

No Comments

No Known Activations