INDEX

Explanations

terms or symbols associated with non-English or encoded content

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With LLAMA3.1-8B @ 24-llamascope-res-32k

Configuration

fnlp/Llama3_1-8B-Base-LXR-8x/Llama3_1-8B-Base-L24R-8x

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

cerebras/SlimPajama-627B

Features

32,768

Data Type

bfloat16

Hook Name

blocks.24.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

lg

-0.21

onth

-0.17

lt

-0.16

rost

-0.15

rypton

-0.15

lero

-0.15

lin

-0.15

odes

-0.15

oms

-0.15

ould

-0.15

POSITIVE LOGITS

eya

0.18

irk

0.18

elow

0.16

iteli

0.16

zer

0.15

enic

0.15

inka

0.15

elez

0.15

 ÑĨ

0.14

edImage

0.14

Activations Density 0.018%

terms or symbols associated with non-English or encoded content

No Comments

No Known Activations