INDEX

Explanations

punctuation and numeric values, particularly around lists and categorization

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

loe

-0.17

lots

-0.15

 Ð¿ÑĢÐ¸Ð´

-0.15

strup

-0.15

erval

-0.15

à¹Īà¸Ńà¸¢

-0.14

 userAgent

-0.14

WithDuration

-0.14

zig

-0.14

eldom

-0.14

POSITIVE LOGITS

etc

0.18

etc

0.17

asin

0.15

up

0.15

oga

0.14

iki

0.14

velle

0.14

IVA

0.14

way

0.14

skin

0.14

Activations Density 0.130%

punctuation and numeric values, particularly around lists and categorization

No Comments

No Known Activations