INDEX

Explanations

image attributions

oai_token-act-pair · gpt-3.5-turbo

repeated phrases or structures emphasizing "of."

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 conduc

-0.87

ashtra

-0.76

resistant

-0.72

 apart

-0.71

ĪĴ

-0.70

 subsequ

-0.70

 attracted

-0.68

 tolerate

-0.68

 evaluates

-0.67

 replaced

-0.67

POSITIVE LOGITS

 STATS

0.85

 Shutterstock

0.81

 Wikimedia

0.73

raphics

0.71

 Freed

0.70

 Shot

0.70

 Leap

0.70

 Media

0.66

 Sierra

0.66

 Wizards

0.66

Activations Density 0.034%

image attributions

repeated phrases or structures emphasizing "of."

No Comments

No Known Activations