INDEX

Explanations

items listed for approval or denial, potentially related to services, products, or coverage

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_fs12288-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_12288

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

12,288

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

 embargo

-0.68

 dere

-0.63

 Register

-0.62

 recording

-0.62

 maxim

-0.62

 imposing

-0.62

 intimidating

-0.60

 sparse

-0.60

 wreck

-0.60

 smugg

-0.59

POSITIVE LOGITS

SourceFile

1.16

*)

1.10

tm

1.07

0.98

edited

0.97

Pg

0.97

+)

0.95

-)

0.94

src

0.93

east

0.92

Activations Density 0.065%

items listed for approval or denial, potentially related to services, products, or coverage

No Comments

No Known Activations