INDEX

Explanations

locations or place names

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_fs24576-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_24576

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

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

yip

-0.72

utterstock

-0.65

itiveness

-0.57

hill

-0.56

manship

-0.55

ibaba

-0.54

Kings

-0.52

spot

-0.52

shit

-0.52

VALUE

-0.51

POSITIVE LOGITS

ace

0.66

astern

0.65

incial

0.62

airo

0.61

adan

0.60

rane

0.60

emporary

0.58

antes

0.57

 Norn

0.56

ãĥī

0.55

Activations Density 0.125%

locations or place names

No Comments

No Known Activations