INDEX

Explanations

words related to modifications or improvements

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

NetMessage

-0.96

chnology

-0.72

Downloadha

-0.68

 shroud

-0.65

 asbestos

-0.65

ttes

-0.64

meal

-0.64

 grease

-0.62

 brace

-0.61

 Mour

-0.61

POSITIVE LOGITS

perial

1.41

ple

1.31

mediately

1.29

plement

1.26

ply

1.26

plementation

1.24

ultane

1.22

agine

1.17

mediate

1.17

itating

1.15

Activations Density 0.485%

words related to modifications or improvements

No Comments

No Known Activations