INDEX

Explanations

elements related to web navigation

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With LLAMA3.1-8B @ 26-llamascope-res-32k

Configuration

fnlp/Llama3_1-8B-Base-LXR-8x/Llama3_1-8B-Base-L26R-8x

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

cerebras/SlimPajama-627B

Features

32,768

Data Type

bfloat16

Hook Name

blocks.26.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

uer

-0.16

ãĥ´

-0.15

ampus

-0.15

emia

-0.15

 GOODMAN

-0.14

 nackte

-0.14

ãĤ¡

-0.13

 ãĥ´

-0.13

udo

-0.13

_overlay

-0.13

POSITIVE LOGITS

 active

0.24

active

0.22

divider

0.20

 divider

0.20

_active

0.20

.separator

0.20

 Active

0.20

.active

0.19

hra

0.19

Divider

0.19

Activations Density 0.028%

elements related to web navigation

No Comments

No Known Activations