INDEX

Explanations

references to spatial directions and placements

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With LLAMA3.1-8B @ 4-llamascope-res-32k

Configuration

fnlp/Llama3_1-8B-Base-LXR-8x/Llama3_1-8B-Base-L4R-8x

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

cerebras/SlimPajama-627B

Features

32,768

Data Type

bfloat16

Hook Name

blocks.4.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

rove

-0.16

ered

-0.14

LOUR

-0.14

ph

-0.14

sey

-0.13

Ø²ÙĪ

-0.13

Ð»ÑĸÐ´

-0.13

illard

-0.13

IVEN

-0.13

Î´Î¹Î±

-0.13

POSITIVE LOGITS

inton

0.16

Pru

0.15

ilters

0.15

istrovstvÃŃ

0.14

byter

0.14

rypto

0.14

mania

0.13

adders

0.13

oldem

0.13

Activations Density 0.020%

references to spatial directions and placements

No Comments

No Known Activations