INDEX

Explanations

programming-related issues and errors

oai_token-act-pair · gpt-4o-mini

Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With LLAMA3.1-8B @ 27-llamascope-res-32k

Configuration

fnlp/Llama3_1-8B-Base-LXR-8x/Llama3_1-8B-Base-L27R-8x

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

cerebras/SlimPajama-627B

Features

32,768

Data Type

bfloat16

Hook Name

blocks.27.hook_resid_post

Hook Layer

Architecture

jumprelu

Context Size

1,024

Dataset

cerebras/SlimPajama-627B

Activation Function

relu

Embeds

IFrame

Link

Negative Logits

hop

-0.14

SP

-0.14

èĲ

-0.14

Î¹ÎºÎ·

-0.14

Ø³ØªØ§ÙĨÛĮ

-0.14

mur

-0.14

IDL

-0.14

analy

-0.14

 Injury

-0.14

-0.13

POSITIVE LOGITS

 raise

0.21

 raising

0.17

<_

0.16

raise

0.16

WRAPPER

0.15

 raised

0.15

 Raised

0.15

 Raise

0.14

ager

0.14

uzzi

0.14

Activations Density 0.003%

programming-related issues and errors

No Known Activations