INDEX

Explanations

names of athletes and sports-related terms

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

orial

-0.85

lishes

-0.81

 Dickens

-0.75

elman

-0.73

arial

-0.73

 Bezos

-0.72

IELD

-0.71

aples

-0.69

SpaceEngineers

-0.68

olphins

-0.67

POSITIVE LOGITS

 rider

0.96

 cyclist

0.94

 riders

0.92

 cycling

0.92

 Cycling

0.91

 doping

0.90

wake

0.85

 cyclists

0.83

bike

0.79

 downhill

0.77

Activations Density 0.052%

names of athletes and sports-related terms

No Comments

No Known Activations