INDEX

Explanations

phrases that connect ideas or provide contrasts in opinions

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

ses

-0.24

etc

-0.18

noch

-0.15

yms

-0.14

ØªÙĨ

-0.14

slideDown

-0.14

inine

-0.14

 blah

-0.14

--

-0.13

–

-0.13

POSITIVE LOGITS

oret

0.28

oretical

0.22

ir

0.20

/-

0.20

!--

0.18

bidden

0.18

ctrine

0.17

ftware

0.17

crease

0.16

phan

0.15

Activations Density 0.124%

phrases that connect ideas or provide contrasts in opinions

No Comments

No Known Activations