INDEX

Explanations

phrases associated with disagreement or differing opinions

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

Fla

-0.64

APTER

-0.62

 Amph

-0.58

 understandably

-0.56

 Tent

-0.55

 Raqqa

-0.54

 Fedora

-0.53

 Ramos

-0.53

 Cage

-0.53

 Blade

-0.53

POSITIVE LOGITS

abouts

1.18

depending

1.05

Else

1.00

 versa

1.00

 thereof

1.00

 altogether

0.94

phans

0.90

 outright

0.88

å§«

0.87

 alternatively

0.85

Activations Density 1.951%

phrases associated with disagreement or differing opinions

No Comments

No Known Activations