INDEX

Explanations

Python code snippets and URLs

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 ......

-0.24

 [...]

-0.23

 amazing

-0.23

 Zerg

-0.22

 Trance

-0.22

AoE

-0.22

catentry

-0.21

 Instruct

-0.21

 Âł

-0.21

Âł

-0.21

POSITIVE LOGITS

etheless

0.23

 Cooke

0.22

treated

0.21

issued

0.21

 Walters

0.21

 Jacobs

0.21

Secretary

0.21

respond

0.21

 outwe

0.20

tains

0.20

Activations Density 23.123%

Python code snippets and URLs

No Comments

No Known Activations