INDEX

Explanations

numerical references, particularly dates and other significant numbers related to historical events and publications

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With LLAMA3.1-8B @ 29-llamascope-res-32k

Configuration

fnlp/Llama3_1-8B-Base-LXR-8x/Llama3_1-8B-Base-L29R-8x

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

cerebras/SlimPajama-627B

Features

32,768

Data Type

bfloat16

Hook Name

blocks.29.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

 Kral

-0.17

¢°

-0.15

APE

-0.15

\Bridge

-0.14

eker

-0.14

hle

-0.14

ÑĩÐ¸Ð½Ð¸

-0.14

kommen

-0.14

Ascii

-0.14

POSITIVE LOGITS

0.35

 Seven

0.29

 seven

0.27

ï¼Ĺ

0.27

 seventh

0.27

-seven

0.26

Seven

0.26

ä¸ĥ

0.25

 Seventh

0.25

à¥Ń

0.25

Activations Density 0.220%

numerical references, particularly dates and other significant numbers related to historical events and publications

No Comments

No Known Activations