INDEX

Explanations

legal or governmental terms and phrases that relate to laws and regulations

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

ĸļ

-0.88

ophon

-0.78

dust

-0.75

Ń·

-0.74

guyen

-0.74

geon

-0.72

emark

-0.71

awar

-0.71

ahar

-0.71

gdala

-0.71

POSITIVE LOGITS

 precautions

0.86

 precaution

0.78

 prerequisite

0.78

 introdu

0.76

 evils

0.74

 prec

0.74

 ingredient

0.72

 steps

0.72

 necessity

0.72

 accompan

0.71

Activations Density 0.033%

legal or governmental terms and phrases that relate to laws and regulations

No Comments

No Known Activations