INDEX

Explanations

countries and their related entities

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

eric

-0.78

onite

-0.72

ufact

-0.72

dylib

-0.71

staking

-0.71

natureconservancy

-0.71

leneck

-0.68

imity

-0.68

ounced

-0.67

netflix

-0.67

POSITIVE LOGITS

®

0.67

Ń

0.67

 Maxwell

0.65

çİĭ

0.65

©

0.65

ĩ

0.65

µ

0.64

 Partnership

0.64

¨

0.62

°

0.61

Activations Density 7.259%

countries and their related entities

No Comments

No Known Activations