INDEX

Explanations

business-related terms, particularly related to customers, employees, and management

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_fs12288-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_12288

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

12,288

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

ļéĨĴ

-0.63

cknow

-0.62

 Pillar

-0.62

MIT

-0.62

 culmination

-0.59

awar

-0.58

NES

-0.57

SourceFile

-0.57

INAL

-0.57

Closure

-0.56

POSITIVE LOGITS

hip

1.49

hips

1.42

 whom

1.10

who

1.07

 willing

1.06

ourced

0.92

folk

0.91

 recruited

0.89

 roaming

0.88

 capable

0.86

Activations Density 0.600%

business-related terms, particularly related to customers, employees, and management

No Comments

No Known Activations