INDEX

Explanations

words related to medical terms or procedures

oai_token-act-pair · gpt-3.5-turbo

specialized terminology related to stratification and population studies

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

 Trek

-0.73

Chuck

-0.71

åĳ

-0.69

 Rowling

-0.67

ctr

-0.67

 serpent

-0.67

RAW

-0.66

urn

-0.65

comings

-0.65

 wolf

-0.64

POSITIVE LOGITS

anian

1.86

atum

1.54

 postp

1.51

lass

1.44

 postponed

1.39

 Caroline

1.38

lasses

1.33

 appliance

1.33

 appliances

1.33

 postpone

1.31

Activations Density 0.059%

words related to medical terms or procedures

specialized terminology related to stratification and population studies

No Comments

No Known Activations