INDEX

Explanations

phrases related to reports or descriptions of events

oai_token-act-pair · gpt-3.5-turbo

phrases that indicate timing or conditions

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

izont

-0.58

ãĥ¼ãĥĨãĤ£

-0.53

ãĤ´ãĥ³

-0.49

ogether

-0.47

gins

-0.46

ãĥ¼ãĥĨ

-0.46

azon

-0.44

ĸļ

-0.44

aturdays

-0.44

ä»£

-0.44

POSITIVE LOGITS

 REUTERS

0.43

GHC

0.43

CBS

0.42

NCT

0.41

 eyewitness

0.40

CVE

0.40

 researchers

0.40

pmwiki

0.38

 astronomers

0.38

 Reuters

0.37

Activations Density 1.195%

phrases related to reports or descriptions of events

phrases that indicate timing or conditions

No Comments

No Known Activations