INDEX

Explanations

references to historical events and political ideologies

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

aton

-0.68

onica

-0.66

 Explosion

-0.66

ral

-0.66

emen

-0.65

atal

-0.65

ODY

-0.65

 Sahara

-0.63

Sem

-0.63

 Xavier

-0.63

POSITIVE LOGITS

etc

1.06

etc

0.94

 circulate

0.92

 uphold

0.90

 engage

0.86

 distribute

0.84

 proclaim

0.84

 collaborate

0.83

 deploy

0.83

 contribute

0.83

Activations Density 0.531%

references to historical events and political ideologies

No Comments

No Known Activations