INDEX

Explanations

references to historical empires and related terms

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

*/(

-0.86

yrics

-0.80

paces

-0.76

electric

-0.74

onto

-0.74

etooth

-0.72

ociate

-0.71

eness

-0.71

matter

-0.70

CAST

-0.69

POSITIVE LOGITS

 Strikes

0.96

 Claud

0.92

perors

0.90

 Augustus

0.86

 Nero

0.84

 empire

0.83

 empires

0.79

perial

0.78

 Napoleon

0.78

 Builder

0.78

Activations Density 0.038%

references to historical empires and related terms

No Comments

No Known Activations