INDEX

Explanations

terms related to roles or roles playing in a particular system or game

oai_token-act-pair · gpt-3.5-turbo

references to roles in gaming contexts

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

bow

-0.74

usters

-0.70

wright

-0.69

aba

-0.69

aho

-0.67

hend

-0.67

 deduct

-0.66

 sweeps

-0.66

 necklace

-0.66

****

-0.63

POSITIVE LOGITS

 Role

3.02

Role

1.91

 Mobil

1.47

 Cors

1.45

 Libre

1.34

Mobil

1.12

 VIDEOS

1.10

role

1.00

rÃ©

0.98

 Ability

0.94

Activations Density 0.053%

terms related to roles or roles playing in a particular system or game

references to roles in gaming contexts

No Comments

No Known Activations