INDEX

Explanations

terms related to architecture and design concepts

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

rof

-0.20

enery

-0.15

-0.14

 estimates

-0.14

 style

-0.14

 Deutsch

-0.14

 frais

-0.14

retch

-0.14

 Expert

-0.13

 stationary

-0.13

POSITIVE LOGITS

 circulation

0.21

 datum

0.20

 volumes

0.19

/site

0.19

 parti

0.18

SITE

0.18

 SITE

0.18

.site

0.18

 cort

0.17

 Site

0.17

Activations Density 0.143%

terms related to architecture and design concepts

No Comments

No Known Activations