INDEX

Explanations

instances of the word "taken", as well as related phrases like "taken out" or "taken hold."

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

holm

-0.67

Cong

-0.65

 Smile

-0.65

 accompanies

-0.64

lich

-0.61

ese

-0.61

 agre

-0.60

tch

-0.59

 livest

-0.59

ingen

-0.57

POSITIVE LOGITS

 advantage

1.31

aways

1.20

 precedence

0.99

 care

0.99

 aback

0.99

 refuge

0.97

 heed

0.94

 precautions

0.91

 liberties

0.88

 pains

0.86

Activations Density 1.926%

instances of the word "taken", as well as related phrases like "taken out" or "taken hold."

No Comments

No Known Activations