INDEX

Explanations

precise descriptions of personal experiences

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

itton

-0.62

ãĥ©ãĥ³

-0.61

accompanied

-0.58

nee

-0.56

 assorted

-0.55

not

-0.55

rather

-0.54

onomy

-0.53

little

-0.52

COLOR

-0.52

POSITIVE LOGITS

 anymore

2.21

nor

2.13

 whatsoever

1.43

nor

1.35

 slightest

1.28

 anything

1.24

 anywhere

1.24

any

1.22

yet

1.13

 anybody

1.11

Activations Density 4.167%

precise descriptions of personal experiences

No Comments

No Known Activations