INDEX

Explanations

specific entities, possibly related to databases, medical terms, or other types of information sources

oai_token-act-pair · gpt-3.5-turbo

the end of the text or document marker

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

emale

-0.67

 horizont

-0.63

 corrid

-0.61

 Appalach

-0.57

pheus

-0.56

 ãĤµãĥ¼ãĥĨãĤ£ãĥ¯ãĥ³

-0.55

 initials

-0.54

ilaterally

-0.52

ornia

-0.51

 vulner

-0.51

POSITIVE LOGITS

lement

0.64

zbollah

0.59

mosp

0.58

Elsa

0.58

NetMessage

0.57

poses

0.57

chwitz

0.57

ANS

0.56

hesda

0.56

olars

0.55

Activations Density 0.302%

specific entities, possibly related to databases, medical terms, or other types of information sources

the end of the text or document marker

No Comments

No Known Activations