INDEX

Explanations

comments or feedback in a written form

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_fs768-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_768

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

768

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

aldehyde

-0.68

ouses

-0.64

esy

-0.60

skirts

-0.59

ANN

-0.58

HG

-0.56

hea

-0.55

lu

-0.55

ike

-0.53

 moreover

-0.53

POSITIVE LOGITS

 relates

0.71

 differs

0.69

 supposed

0.69

phr

0.67

ribed

0.67

 geographically

0.66

enth

0.64

 relate

0.64

icz

0.63

 unfolded

0.63

Activations Density 3.785%

comments or feedback in a written form

No Comments

No Known Activations