INDEX

Explanations

references to doors in various contexts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With LLAMA3.1-8B @ 29-llamascope-res-32k

Configuration

fnlp/Llama3_1-8B-Base-LXR-8x/Llama3_1-8B-Base-L29R-8x

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

cerebras/SlimPajama-627B

Features

32,768

Data Type

bfloat16

Hook Name

blocks.29.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

dic

-0.18

adele

-0.17

vrier

-0.16

tl

-0.16

tank

-0.16

tlement

-0.15

tin

-0.15

ertools

-0.15

kar

-0.15

kle

-0.15

POSITIVE LOGITS

ways

0.41

bell

0.39

frame

0.30

steps

0.27

/window

0.26

keeper

0.26

 knob

0.26

handle

0.25

frames

0.25

step

0.24

Activations Density 0.030%

references to doors in various contexts

No Comments

No Known Activations