INDEX

Explanations

references to hats or hat-related concepts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With LLAMA3.1-8B @ 24-llamascope-res-32k

Configuration

fnlp/Llama3_1-8B-Base-LXR-8x/Llama3_1-8B-Base-L24R-8x

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

cerebras/SlimPajama-627B

Features

32,768

Data Type

bfloat16

Hook Name

blocks.24.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

etto

-0.18

quat

-0.17

eron

-0.17

icine

-0.16

eros

-0.16

tank

-0.15

usercontent

-0.15

bett

-0.15

tiles

-0.15

awl

-0.15

POSITIVE LOGITS

field

0.21

RED

0.20

red

0.20

 trick

0.20

rack

0.19

ting

0.19

 Trick

0.19

 tricks

0.19

reds

0.18

maker

0.18

Activations Density 0.009%

references to hats or hat-related concepts

No Comments

No Known Activations