INDEX

Explanations

mixed opinions and uncertainties in answers given to specific questions

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

icipated

-0.64

drawn

-0.57

 Roll

-0.56

ership

-0.56

ipal

-0.56

riers

-0.55

isin

-0.54

 buildup

-0.54

 Skydragon

-0.54

Merit

-0.53

POSITIVE LOGITS

yes

1.18

 Nope

1.10

YES

1.08

YES

1.03

yes

1.02

 answer

1.01

Yes

1.00

Answer

0.94

Absolutely

0.91

 anyway

0.91

Activations Density 2.341%

mixed opinions and uncertainties in answers given to specific questions

No Comments

No Known Activations