INDEX

Explanations

phrases indicating generic choices or possibilities

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

livious

-0.77

undown

-0.74

ļéĨĴ

-0.73

ridor

-0.68

ĸļ

-0.67

eworthy

-0.67

 Izan

-0.63

aza

-0.62

seless

-0.61

 appre

-0.60

POSITIVE LOGITS

 decides

0.89

 chooses

0.88

 decide

0.86

pires

0.86

may

0.80

 choose

0.71

 might

0.69

 happens

0.68

 whatsoever

0.67

 winds

0.66

Activations Density 0.153%

phrases indicating generic choices or possibilities

No Comments

No Known Activations