INDEX

Explanations

phrases related to news articles, incidents, or reports

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_fs1536-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_1536

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

1,536

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

anwhile

-0.81

 UNCLASSIFIED

-0.79

ogether

-0.73

 versa

-0.73

_.

-0.72

 encour

-0.71

theless

-0.71

[*

-0.70

ļéĨĴ

-0.67

pse

-0.67

POSITIVE LOGITS

 Belfast

0.88

 âĢº

0.83

 Donald

0.79

 Transgender

0.72

 Hillary

0.70

 Hearthstone

0.69

 âĢİ

0.68

][

0.66

 Updated

0.65

 cryptocurrency

0.65

Activations Density 4.217%

phrases related to news articles, incidents, or reports

No Comments

No Known Activations