INDEX

Explanations

specific names or titles relevant to characters, places, or organizations

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

Negative Logits

ÑĢÑİ

-0.16

bgcolor

-0.15

oka

-0.15

folio

-0.14

ÑģÑı

-0.14

_CTL

-0.14

ationToken

-0.14

 outnumber

-0.13

nyder

-0.13

 suÃ§

-0.13

POSITIVE LOGITS

mayan

0.16

æ¢°

0.16

uhl

0.14

Ð°Ð´Ñĥ

0.14

rello

0.14

ãģŁãģĹ

0.14

ãĥ¬ãĥĥãĥĪ

0.14

hee

0.14

alli

0.14

orer

0.13

Activations Density 0.007%

specific names or titles relevant to characters, places, or organizations

No Known Activations