INDEX

Explanations

HTML class and ID attributes that indicate structure or layout elements

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

 strat

-0.15

claimer

-0.14

.opensource

-0.14

::*

-0.14

 AydÄ±n

-0.14

 hell

-0.14

emin

-0.14

.toArray

-0.13

inar

-0.13

ORAGE

-0.13

POSITIVE LOGITS

="

0.22

 Nolan

0.17

ê±°

0.15

ä¼¼

0.14

Sto

0.14

icro

0.14

ADDE

0.14

='

0.14

eshire

0.14

Activations Density 0.010%

HTML class and ID attributes that indicate structure or layout elements

No Comments

No Known Activations