INDEX

Explanations

information related to specific topics and points in a 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_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

 purported

-0.72

ELD

-0.70

ynthesis

-0.69

erial

-0.63

 purportedly

-0.60

CVE

-0.60

 rendered

-0.58

Various

-0.57

impl

-0.56

terday

-0.56

POSITIVE LOGITS

 yourself

1.64

 yourselves

1.58

 Yourself

1.47

 your

1.17

 beware

1.07

 wisely

1.02

 YOUR

1.01

ichever

0.99

 Your

0.96

 responsibly

0.96

Activations Density 5.722%

information related to specific topics and points in a text

No Comments

No Known Activations