INDEX

Explanations

instances of time-related information or sequences

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With LLAMA3.1-8B @ 18-llamascope-res-32k

Configuration

fnlp/Llama3_1-8B-Base-LXR-8x/Llama3_1-8B-Base-L18R-8x

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

cerebras/SlimPajama-627B

Features

32,768

Data Type

bfloat16

Hook Name

blocks.18.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

irut

-0.15

çĦ¼

-0.14

uve

-0.14

RATION

-0.14

ó

-0.14

aney

-0.14

ABEL

-0.13

antino

-0.13

otten

-0.13

POSITIVE LOGITS

renom

0.17

al

0.17

Al

0.16

amba

0.15

guard

0.15

astr

0.15

ing

0.15

 Sala

0.14

Unload

0.14

 Gard

0.14

Activations Density 0.016%

instances of time-related information or sequences

No Comments

No Known Activations