INDEX

Explanations

phrases related to scientific research and organizations

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

rises

-0.73

reads

-0.70

depending

-0.69

adow

-0.69

ãĥİ

-0.68

hend

-0.68

azing

-0.68

stroke

-0.67

slice

-0.66

inventoryQuantity

-0.64

POSITIVE LOGITS

1.10

("

1.05

 Commission

1.03

 Directorate

1.02

 Association

1.01

 Foundation

1.01

Ltd

0.98

 Board

0.96

 Consortium

0.94

Act

0.93

Activations Density 4.030%

phrases related to scientific research and organizations

No Comments

No Known Activations