INDEX

Explanations

sentences related to news articles featuring announcements and interactions between notable individuals

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

ichick

-0.89

ãĥ¼ãĥĨãĤ£

-0.74

²

-0.68

ãĤ©

-0.65

ãĥ¼ãĥĨ

-0.63

 attRot

-0.62

 monet

-0.62

ãĤ¡

-0.61

yt

-0.61

 endeavor

-0.60

POSITIVE LOGITS

Scientists

0.88

 Britons

0.84

Published

0.83

Britain

0.83

Sorry

0.82

Updated

0.82

Former

0.82

Actor

0.80

Thousands

0.77

 Latest

0.76

Activations Density 0.134%

sentences related to news articles featuring announcements and interactions between notable individuals

No Comments

No Known Activations