INDEX

Explanations

references to news articles or reports

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_fs24576-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_24576

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

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

 discrim

-0.64

 expire

-0.64

 fung

-0.63

 orche

-0.63

 majesty

-0.62

 registry

-0.62

 plants

-0.62

 ineffective

-0.61

 hops

-0.60

 chop

-0.60

POSITIVE LOGITS

Reuters

1.30

Thom

1.19

CBS

1.05

AFP

1.00

CNN

0.97

Associated

0.95

CN

0.92

credit

0.92

AP

0.90

zbollah

0.89

Activations Density 0.030%

references to news articles or reports

No Comments

No Known Activations