INDEX

Explanations

words related to mathematical equations and terms

oai_token-act-pair · gpt-3.5-turbo

quantitative or mathematical expressions

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

rosse

-0.64

oppers

-0.55

ependence

-0.49

oeuv

-0.49

inders

-0.48

 belie

-0.48

ummer

-0.44

 conserve

-0.44

turn

-0.43

inge

-0.43

POSITIVE LOGITS

 acronym

0.72

 abbre

0.60

REDACTED

0.59

ITION

0.57

 Vulkan

0.56

£ı

0.54

 Digital

0.52

Ö¼

0.52

VK

0.52

ãĤ±

0.51

Activations Density 1.944%

words related to mathematical equations and terms

quantitative or mathematical expressions

No Comments

No Known Activations