INDEX

Explanations

proper nouns related to politics and news

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_fs768-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_768

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

768

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

olulu

-0.73

FU

-0.68

 Krish

-0.67

ories

-0.66

 Beir

-0.66

 Celt

-0.65

 Tamil

-0.64

 Pixie

-0.64

 Artemis

-0.62

raints

-0.61

POSITIVE LOGITS

Jr

1.14

's

1.06

Sr

0.97

 supporters

0.93

 himself

0.93

 Jinping

0.92

 confid

0.89

omics

0.89

 supporter

0.86

III

0.85

Activations Density 3.594%

proper nouns related to politics and news

No Comments

No Known Activations