INDEX

Explanations

phrases related to advice and personal experiences

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

å£«

-0.64

 Continental

-0.61

 Penguin

-0.59

 SetTextColor

-0.58

 Gaul

-0.57

 Crus

-0.56

 Bernstein

-0.55

 Mobil

-0.55

 Distribution

-0.54

ombat

-0.54

POSITIVE LOGITS

selves

1.17

theless

1.10

 gonna

0.95

ready

0.94

initely

0.93

been

0.87

actly

0.83

pecially

0.82

self

0.81

etheless

0.81

Activations Density 2.706%

phrases related to advice and personal experiences

No Comments

No Known Activations