INDEX

Explanations

words related to legal terms

oai_token-act-pair · gpt-3.5-turbo

phrases related to unusual or non-standard terms and names

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 2-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.2.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.2.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

 CLSID

-0.43

 examiner

-0.41

 reused

-0.40

 host

-0.40

 guid

-0.39

 preserved

-0.39

 Dominican

-0.39

 transition

-0.39

 subdiv

-0.38

 starter

-0.38

POSITIVE LOGITS

etheless

0.62

oche

0.55

astic

0.55

terness

0.55

asma

0.54

lement

0.53

theless

0.53

ove

0.53

ffect

0.52

asy

0.52

Activations Density 1.718%

words related to legal terms

phrases related to unusual or non-standard terms and names

No Comments

No Known Activations