INDEX

Explanations

names or titles related to individuals or groups of people

oai_token-act-pair · gpt-3.5-turbo

names of characters and locations in narratives

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

OOL

-0.90

chat

-0.75

ibaba

-0.74

STD

-0.71

advertisement

-0.71

oteric

-0.71

#$

-0.70

ko

-0.69

otype

-0.67

 skiing

-0.66

POSITIVE LOGITS

 Lords

1.70

 Cris

1.48

 Garn

1.34

 Madd

1.34

 Cass

1.33

enda

1.21

 Brut

1.18

 Haven

1.10

 Prosper

1.09

 Trem

1.07

Activations Density 0.076%

names or titles related to individuals or groups of people

names of characters and locations in narratives

No Comments

No Known Activations