INDEX

Explanations

phrases related to accuracy and correctness in processes or information

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With LLAMA3.1-8B @ 21-llamascope-res-32k

Configuration

fnlp/Llama3_1-8B-Base-LXR-8x/Llama3_1-8B-Base-L21R-8x

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

cerebras/SlimPajama-627B

Features

32,768

Data Type

bfloat16

Hook Name

blocks.21.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

ãĥ³ãĤº

-0.16

ongoose

-0.15

onet

-0.15

 Progress

-0.15

ìľ¨

-0.15

-0.14

 credential

-0.14

_PROGRESS

-0.14

MOOTH

-0.13

etur

-0.13

POSITIVE LOGITS

 wrong

0.42

 correct

0.39

wrong

0.36

 right

0.35

 Wrong

0.34

 WRONG

0.34

Wrong

0.33

correct

0.31

right

0.30

 correctly

0.29

Activations Density 0.059%

phrases related to accuracy and correctness in processes or information

No Comments

No Known Activations