INDEX

Explanations

political and militaristic terms

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_fs1536-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_1536

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

1,536

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

oret

-0.55

Aus

-0.48

 mend

-0.46

 depends

-0.45

 Philly

-0.45

utters

-0.44

 Oliv

-0.44

 Leaf

-0.44

 biome

-0.44

IPM

-0.43

POSITIVE LOGITS

 respectively

1.12

etheless

0.85

*.

0.67

umbered

0.64

 guiName

0.64

NetMessage

0.63

>.

0.60

.).

0.59

.</

0.59

inguished

0.58

Activations Density 9.488%

political and militaristic terms

No Comments

No Known Activations