INDEX

Explanations

Twitter usernames

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_fs3072-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_3072

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

3,072

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

 LIMITED

-0.63

 tense

-0.63

 clinch

-0.62

 bindings

-0.62

 ACTIONS

-0.62

 striking

-0.61

 pressing

-0.61

 "âĢ¦

-0.61

 Ninth

-0.61

Vi

-0.60

POSITIVE LOGITS

1.22

official

1.08

CBC

1.00

Writ

0.98

reports

0.98

Reports

0.98

DonaldTrump

0.93

veyard

0.91

gamer

0.91

GoldMagikarp

0.89

Activations Density 0.639%

Twitter usernames

No Comments

No Known Activations