INDEX

Explanations

terms related to laws, regulations, and judgments

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

usalem

-0.63

©¶æ

-0.62

chance

-0.57

sten

-0.56

href

-0.56

ravings

-0.56

ighth

-0.56

agine

-0.54

çīĪ

-0.53

asio

-0.53

POSITIVE LOGITS

 enough

1.32

 insofar

1.05

enough

1.02

 compared

0.91

 because

0.90

ly

0.90

isable

0.90

 Enough

0.88

 nonetheless

0.86

 anyway

0.85

Activations Density 4.770%

terms related to laws, regulations, and judgments

No Comments

No Known Activations