INDEX

Explanations

references to reports and investigations

oai_token-act-pair · gpt-3.5-turbo

occurrences of the word "report" and related terms

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

east

-0.61

conservancy

-0.60

HAM

-0.57

ansk

-0.57

 Denis

-0.56

frey

-0.56

 Mechdragon

-0.54

amp

-0.53

TON

-0.53

 radial

-0.53

POSITIVE LOGITS

 commissioned

1.00

 titled

0.95

 detailing

0.95

card

0.91

 card

0.89

 declass

0.88

orial

0.85

 entitled

0.85

 documenting

0.85

recomm

0.84

Activations Density 0.098%

references to reports and investigations

occurrences of the word "report" and related terms

No Comments

No Known Activations