INDEX

Explanations

references to walls and their decorations or placements

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

ìĭł

-0.19

atica

-0.16

-0.15

uchar

-0.15

marsh

-0.14

éĥ

-0.14

strup

-0.14

seau

-0.14

ä¼ı

-0.14

lain

-0.14

POSITIVE LOGITS

 wall

0.74

 walls

0.70

wall

0.62

-wall

0.60

 WALL

0.59

 Wall

0.56

.wall

0.56

walls

0.54

Wall

0.54

å¢Ļ

0.54

Activations Density 0.159%

references to walls and their decorations or placements

No Comments

No Known Activations