INDEX

Explanations

quantifiers and references to groups or categories

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

oucher

-0.16

uel

-0.15

achine

-0.14

vie

-0.14

doctype

-0.14

ui

-0.13

uren

-0.13

Wir

-0.13

ock

-0.13

eer

-0.13

POSITIVE LOGITS

 them

0.23

 which

0.23

 ones

0.23

them

0.21

 ÐºÐ¾ÑĤÐ¾ÑĢÑĭÑħ

0.20

maal

0.20

 ÐºÐ¾ÑĤÐ¾ÑĢÐ¾Ð¹

0.19

 cuales

0.19

 with

0.19

 being

0.17

Activations Density 0.053%

quantifiers and references to groups or categories

No Comments

No Known Activations