INDEX

Explanations

mentions of the term "Deep" with different emphasis

oai_token-act-pair · gpt-3.5-turbo

references to the Deepwater Horizon incident

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

cules

-0.87

uably

-0.77

ICAN

-0.77

 Speedway

-0.72

ä»£

-0.68

Ń·

-0.67

perty

-0.66

icans

-0.66

 grace

-0.63

ICA

-0.62

POSITIVE LOGITS

water

0.96

ly

0.85

 vein

0.84

ness

0.80

lings

0.76

 penetration

0.73

ened

0.73

croft

0.72

redo

0.71

est

0.70

Activations Density 0.012%

mentions of the term "Deep" with different emphasis

references to the Deepwater Horizon incident

No Comments

No Known Activations