INDEX

Explanations

names of individuals

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

 ultras

-0.81

 Galileo

-0.69

 Duterte

-0.68

 Kazakh

-0.67

 Sakuya

-0.67

 Kazakhstan

-0.66

 unfavorable

-0.65

 Croatian

-0.64

 Bulgarian

-0.64

 globalization

-0.63

POSITIVE LOGITS

ford

1.23

worth

1.18

ley

1.18

burgh

1.17

field

1.15

beck

1.15

ridge

1.14

cliffe

1.12

herty

1.11

ney

1.11

Activations Density 0.276%

names of individuals

No Comments

No Known Activations