INDEX

Explanations

concepts related to concealment or deception

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With LLAMA3.1-8B @ 21-llamascope-res-32k

Configuration

fnlp/Llama3_1-8B-Base-LXR-8x/Llama3_1-8B-Base-L21R-8x

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

cerebras/SlimPajama-627B

Features

32,768

Data Type

bfloat16

Hook Name

blocks.21.hook_resid_post

Hook Layer

Architecture

jumprelu

Context Size

1,024

Dataset

cerebras/SlimPajama-627B

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

ITUDE

-0.19

ois

-0.15

 Giov

-0.15

illard

-0.15

elles

-0.14

Ð¾Ð±Ñĭ

-0.14

é©

-0.14

seed

-0.14

oner

-0.13

PCS

-0.13

POSITIVE LOGITS

 behind

0.29

away

0.28

 hide

0.23

 Behind

0.23

 away

0.22

 hides

0.22

ously

0.22

.hide

0.20

 hiding

0.20

Behind

0.20

Activations Density 0.032%

concepts related to concealment or deception

No Comments

No Known Activations