INDEX

Explanations

contractions and affirmative statements about unspecified subjects

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

ollo

-0.15

usch

-0.14

/octet

-0.14

parsers

-0.14

çĶ£

-0.13

assin

-0.13

lique

-0.13

 gá»ĳc

-0.13

agh

-0.13

licken

-0.13

POSITIVE LOGITS

 remains

0.26

 appears

0.25

 appear

0.23

 marks

0.21

 remain

0.20

 Appears

0.19

zel

0.19

 turns

0.18

appe

0.18

 looks

0.17

Activations Density 0.083%

contractions and affirmative statements about unspecified subjects

No Comments

No Known Activations