INDEX

Explanations

quotations within text

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_fs1536-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_1536

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

1,536

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

 parting

-0.80

 pudding

-0.78

 favor

-0.77

 secret

-0.70

 dispar

-0.70

 classified

-0.70

 landslide

-0.70

 bunk

-0.69

 dominate

-0.69

 friendly

-0.69

POSITIVE LOGITS

It

1.48

We

1.46

They

1.44

There

1.42

Because

1.39

Especially

1.37

Sometimes

1.36

1.35

But

1.35

Obviously

1.34

Activations Density 1.058%

quotations within text

No Comments

No Known Activations