INDEX

Explanations

negative statements expressing disagreement or refusal

oai_token-act-pair · gpt-3.5-turbo Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res_fs24576-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_24576

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

torch.float32

Hook Point

blocks.8.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

å¥

-0.75

 Might

-0.70

éĹĺ

-0.69

æ©

-0.68

PsyNetMessage

-0.66

çīĪ

-0.65

Writer

-0.65

 Tours

-0.64

 Tens

-0.63

might

-0.63

POSITIVE LOGITS

 necessarily

1.38

icable

1.32

icably

1.29

eworthy

1.03

yet

1.00

epad

1.00

hin

1.00

 exactly

0.96

 allowed

0.96

orious

0.96

Activations Density 0.124%

negative statements expressing disagreement or refusal

No Comments

No Known Activations