INDEX

Explanations

numbers and percentages

oai_token-act-pair · gpt-3.5-turbo

phrases indicative of recommendations or suggestions

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

 Snowden

-0.56

 someday

-0.55

last

-0.53

 Clapper

-0.52

 sooner

-0.52

 Guant

-0.51

olars

-0.49

 sometime

-0.47

insk

-0.47

CVE

-0.47

POSITIVE LOGITS

 thereof

0.61

him

0.59

edo

0.54

emale

0.51

ãĥ¼ãĥ«

0.48

 versa

0.48

SourceFile

0.47

 confir

0.46

alian

0.46

ya

0.46

Activations Density 0.789%

numbers and percentages

phrases indicative of recommendations or suggestions

No Comments

No Known Activations