INDEX

Explanations

numerical values specified in a comparison

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 nude

-0.70

 masked

-0.68

 sensitive

-0.67

 offended

-0.67

 slick

-0.67

 wellness

-0.67

 outreach

-0.67

 timely

-0.66

vice

-0.65

 invaluable

-0.64

POSITIVE LOGITS

 Either

1.40

 Both

1.35

 Therefore

1.17

 Whatever

1.16

 Neither

1.16

 Again

1.15

 Regardless

1.15

 Thus

1.14

 Together

1.13

 Hence

1.13

Activations Density 0.545%

numerical values specified in a comparison

No Comments

No Known Activations