INDEX

Explanations

names of people

oai_token-act-pair · gpt-3.5-turbo

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

Nar

-0.57

 Breaker

-0.57

ãĤ¤

-0.55

ÃįÃį

-0.55

Ø³

-0.53

PDATE

-0.52

 Newtown

-0.52

izoph

-0.52

 MDMA

-0.51

 HUGE

-0.51

POSITIVE LOGITS

's

1.27

 himself

1.22

 testified

0.96

 herself

0.95

 Productions

0.93

 wrote

0.91

 remembers

0.89

 admits

0.86

 presided

0.85

 recalls

0.84

Activations Density 0.229%

names of people

No Comments

No Known Activations