INDEX

Explanations

numbers and punctuation marks

oai_token-act-pair · gpt-3.5-turbo

structured elements commonly found in legal documents and discussions

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

CLR

-0.72

Sounds

-0.69

 guiActiveUnfocused

-0.68

 Gleaming

-0.67

è¦

-0.67

uala

-0.67

UID

-0.66

emort

-0.66

 Tribunal

-0.65

 Cosponsors

-0.64

POSITIVE LOGITS

emonic

0.72

dx

0.71

mma

0.70

dm

0.69

politics

0.69

kl

0.68

stem

0.68

example

0.66

facebook

0.66

rc

0.65

Activations Density 0.248%

numbers and punctuation marks

structured elements commonly found in legal documents and discussions

No Comments

No Known Activations