INDEX

Explanations

names of airlines and flight-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

arios

-0.75

idency

-0.67

ebook

-0.66

kai

-0.65

hunt

-0.65

icle

-0.64

otte

-0.64

olithic

-0.64

isite

-0.63

 Stim

-0.63

POSITIVE LOGITS

 Airlines

1.09

 passenger

1.04

 passengers

0.97

 flights

0.91

 airlines

0.90

 airline

0.89

 flight

0.87

 Flight

0.86

Pass

0.83

 hangar

0.83

Activations Density 0.044%

names of airlines and flight-related terms

No Comments

No Known Activations