INDEX

Explanations

intensifiers or modifiers that emphasize a degree of quality or extent

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

ict

-0.15

 mere

-0.15

ÑĭÑħ

-0.15

 Better

-0.15

irk

-0.15

 better

-0.14

better

-0.14

.XR

-0.14

 âĹ

-0.14

 Really

-0.14

POSITIVE LOGITS

 much

0.35

much

0.34

 Much

0.27

Much

0.25

 MUCH

0.24

yyyy

0.22

 seldom

0.21

yy

0.21

 rarely

0.21

DOC

0.20

Activations Density 0.071%

intensifiers or modifiers that emphasize a degree of quality or extent

No Comments

No Known Activations