INDEX

Explanations

phrases related to factual information and statements

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

 Reef

-0.70

 Weston

-0.66

 Entered

-0.58

Kau

-0.58

KS

-0.57

 Survivors

-0.56

 Shepard

-0.56

 Ninth

-0.56

 Allied

-0.56

 Proceedings

-0.56

POSITIVE LOGITS

theless

1.12

usterity

1.07

terday

1.02

'?

0.99

\)

0.99

Â»

0.95

0.94

%"

0.93

lihood

0.91

estine

0.88

Activations Density 0.312%

phrases related to factual information and statements

No Comments

No Known Activations