INDEX

Explanations

technical terms related to technology and weaponry

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

theless

-0.69

train

-0.62

inventoryQuantity

-0.60

 exha

-0.59

ĺħ

-0.59

 condem

-0.58

lax

-0.57

nesota

-0.56

wagen

-0.56

 ment

-0.56

POSITIVE LOGITS

ribute

1.33

olerance

1.27

ractor

1.23

ributes

1.21

ARGET

1.21

itles

1.20

olkien

1.19

itled

1.15

ournaments

1.15

ribut

1.14

Activations Density 3.788%

technical terms related to technology and weaponry

No Comments

No Known Activations