INDEX

Explanations

information related to historical inventions and figures

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

autions

-0.78

charges

-0.78

affected

-0.73

ef

-0.73

Alert

-0.72

inventoryQuantity

-0.72

wcsstore

-0.72

Prov

-0.71

liga

-0.69

 disclose

-0.68

POSITIVE LOGITS

 modern

1.03

 proto

0.78

 civilization

0.77

cyclopedia

0.75

 Civilization

0.74

 Encyclopedia

0.74

 famous

0.73

 Macintosh

0.73

 Golem

0.73

 Modern

0.72

Activations Density 0.317%

information related to historical inventions and figures

No Comments

No Known Activations