INDEX

Explanations

words related to legal or criminal activities

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

 reckoning

-0.69

 Mara

-0.59

 Benz

-0.58

 Saban

-0.58

 Brah

-0.57

 Cheong

-0.57

yip

-0.56

 tremend

-0.56

WARD

-0.55

 Ragnarok

-0.55

POSITIVE LOGITS

aneous

0.99

ciation

0.94

incial

0.92

etary

0.90

rency

0.88

ctive

0.88

ciating

0.84

cled

0.83

ential

0.82

acion

0.81

Activations Density 4.289%

words related to legal or criminal activities

No Comments

No Known Activations