INDEX

Explanations

locations and government-related terms, especially related to the United States government and its 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

bably

-0.73

*/(

-0.67

odor

-0.63

ebook

-0.61

aird

-0.61

ients

-0.60

errors

-0.60

plates

-0.59

lets

-0.59

sym

-0.59

POSITIVE LOGITS

 Embassy

1.13

 Postal

1.11

 embassy

1.00

 Geological

0.99

 Dollar

0.98

 Airlines

0.98

 Airways

0.97

AAF

0.95

 Army

0.91

 Treasury

0.89

Activations Density 2.032%

locations and government-related terms, especially related to the United States government and its activities

No Comments

No Known Activations