INDEX

Explanations

names of people or entities

oai_token-act-pair · gpt-3.5-turbo

proper nouns related to specific individuals or characters

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

Virginia

-0.74

 decimal

-0.73

 bunker

-0.70

wall

-0.67

iances

-0.67

dot

-0.65

iscal

-0.64

iff

-0.63

mont

-0.62

 Malcolm

-0.61

POSITIVE LOGITS

Sak

4.00

 Shak

1.52

Nak

1.52

Suk

1.49

Sek

1.48

Rak

1.47

Tak

1.40

 Nish

1.38

 Tanaka

1.37

 Kats

1.33

Activations Density 0.025%

names of people or entities

proper nouns related to specific individuals or characters

No Comments

No Known Activations