INDEX

Explanations

phrases related to architectural features or elements

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

ensa

-0.17

ÏĩÎµÎ¹

-0.17

ensch

-0.16

kins

-0.15

Translator

-0.15

erland

-0.15

 congen

-0.15

bull

-0.15

indre

-0.14

cap

-0.14

POSITIVE LOGITS

acam

0.16

 Hamp

0.16

ican

0.15

(pc

0.15

urv

0.14

ppo

0.14

tie

0.14

PC

0.14

 ÎļÎ±Ïģ

0.14

chester

0.14

Activations Density 0.029%

phrases related to architectural features or elements

No Comments

No Known Activations