INDEX

Explanations

phrases related to player actions or attributes

oai_token-act-pair · gpt-3.5-turbo

references to players and viewers in gaming or interactive contexts

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

ories

-0.76

Ø§ÙĦ

-0.76

hell

-0.74

arus

-0.73

equality

-0.73

?????-

-0.72

hots

-0.71

 Cosponsors

-0.69

lap

-0.69

 Delicious

-0.68

POSITIVE LOGITS

's

0.91

 perce

0.89

ÃŃs

0.83

 chooses

0.80

Â´

0.78

 desired

0.75

osphere

0.74

 compartment

0.74

 immersed

0.73

 verbally

0.72

Activations Density 0.205%

phrases related to player actions or attributes

references to players and viewers in gaming or interactive contexts

No Comments

No Known Activations