INDEX

Explanations

instances of the word "one" associated with numerical values, establishing importance or comparison

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

ooks

-0.82

hips

-0.77

ories

-0.75

inders

-0.73

respective

-0.69

osponsors

-0.68

ourn

-0.63

emies

-0.62

folk

-0.62

actions

-0.61

POSITIVE LOGITS

 hundred

1.19

 Hundred

1.10

 thousand

0.99

 thing

0.92

 sided

0.90

 dimensional

0.87

 month

0.83

 Piece

0.83

 Thousand

0.83

Shot

0.81

Activations Density 1.813%

instances of the word "one" associated with numerical values, establishing importance or comparison

No Comments

No Known Activations