INDEX

Explanations

references to lost, hidden, or unreleased content or materials

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With LLAMA3.1-8B @ 9-llamascope-res-32k

Configuration

fnlp/Llama3_1-8B-Base-LXR-8x/Llama3_1-8B-Base-L9R-8x

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

cerebras/SlimPajama-627B

Features

32,768

Data Type

bfloat16

Hook Name

blocks.9.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

bildung

-0.15

<Props

-0.14

WRAPPER

-0.14

åĿĤ

-0.14

RESS

-0.14

Mess

-0.14

/goto

-0.14

mess

-0.14

finder

-0.14

æµ¦

-0.14

POSITIVE LOGITS

 side

0.18

ashboard

0.17

 Side

0.16

Thi

0.16

 hidden

0.16

 buried

0.16

_side

0.15

çŀ

0.15

osc

0.15

 recess

0.15

Activations Density 0.101%

references to lost, hidden, or unreleased content or materials

No Comments

No Known Activations