INDEX

Explanations

evaluative language regarding quality and performance

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With LLAMA3.1-8B @ 20-llamascope-res-32k

Configuration

fnlp/Llama3_1-8B-Base-LXR-8x/Llama3_1-8B-Base-L20R-8x

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

cerebras/SlimPajama-627B

Features

32,768

Data Type

bfloat16

Hook Name

blocks.20.hook_resid_post

Hook Layer

Architecture

jumprelu

Context Size

1,024

Dataset

cerebras/SlimPajama-627B

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 goodness

-0.20

ingham

-0.17

 greatness

-0.16

 excellence

-0.15

 Excellence

-0.15

imler

-0.15

.sha

-0.14

ONA

-0.14

igo

-0.14

POSITIVE LOGITS

-quality

0.24

quality

0.18

/fast

0.17

 quality

0.17

/loose

0.16

byss

0.16

 prognosis

0.15

à¤¤à¤®

0.15

å¤ªéĥİ

0.15

 íĴĪ

0.15

Activations Density 0.133%

evaluative language regarding quality and performance

No Comments

No Known Activations