INDEX

Explanations

phrases related to document formatting and editing

oai_token-act-pair · gpt-3.5-turbo

phrases indicating conditional or qualifying statements about content variation and form

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

ware

-0.76

orship

-0.64

hower

-0.64

chnology

-0.61

gaard

-0.61

onite

-0.58

geist

-0.58

xia

-0.57

glers

-0.57

ologies

-0.56

POSITIVE LOGITS

eter

0.52

0.51

Eng

0.50

ali

0.48

ini

0.47

LI

0.45

ILL

0.44

eno

0.44

est

0.44

idi

0.43

Activations Density 0.248%

phrases related to document formatting and editing

phrases indicating conditional or qualifying statements about content variation and form

No Comments

No Known Activations