INDEX

Explanations

the word "so" with varying degrees of emphasis

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_fs3072-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_3072

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

3,072

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

theless

-0.72

works

-0.61

work

-0.58

 expectancy

-0.56

 Purg

-0.55

amac

-0.52

 Flavoring

-0.52

 Halls

-0.52

 eviction

-0.51

 Mens

-0.51

POSITIVE LOGITS

bered

1.31

oths

1.26

apy

1.21

othes

1.18

oooo

1.11

othe

1.11

ooo

1.09

oner

1.05

far

0.96

aps

0.94

Activations Density 0.255%

the word "so" with varying degrees of emphasis

No Comments

No Known Activations