INDEX

Explanations

scientific terms and concepts, especially related to biology, research, and technical systems

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

 Encyclopedia

-0.68

 Elves

-0.66

lihood

-0.65

 Kubrick

-0.63

SourceFile

-0.61

 Icelandic

-0.61

bed

-0.61

AVG

-0.60

 Albion

-0.59

WAY

-0.59

POSITIVE LOGITS

otional

1.41

phasis

1.35

otions

1.27

otion

1.27

peror

1.26

igrants

1.25

igrant

1.25

igration

1.23

perors

1.21

issions

1.21

Activations Density 0.576%

scientific terms and concepts, especially related to biology, research, and technical systems

No Comments

No Known Activations