INDEX

Explanations

phrases related to scenarios and events

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 alike

-0.84

 interviewed

-0.68

idents

-0.67

 impressions

-0.67

 diets

-0.67

casters

-0.64

 cues

-0.62

 intakes

-0.62

 foregoing

-0.62

chell

-0.61

POSITIVE LOGITS

 which

1.20

efficiency

0.95

 lieu

0.94

which

0.87

 Which

0.85

accessible

0.80

icial

0.80

patient

0.78

 whose

0.76

activated

0.76

Activations Density 0.313%

phrases related to scenarios and events

No Comments

No Known Activations