INDEX

Explanations

phrases related to superlative expressions of quality or status

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

 predecessors

-0.16

çļĦæĺ¯

-0.15

anna

-0.15

nie

-0.15

ieval

-0.14

 nearest

-0.14

 partners

-0.14

unky

-0.14

 Ashe

-0.14

POSITIVE LOGITS

 largest

0.18

 leading

0.17

é¾Ħ

0.17

larg

0.17

Fin

0.16

ryo

0.15

andest

0.15

 most

0.15

 Largest

0.15

 premier

0.14

Activations Density 0.048%

phrases related to superlative expressions of quality or status

No Comments

No Known Activations