INDEX

Explanations

terms related to scoring and evaluations

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

ekil

-0.17

igm

-0.17

ector

-0.17

rud

-0.16

-webpack

-0.15

oral

-0.15

orp

-0.15

shade

-0.15

dehyde

-0.15

rem

-0.15

POSITIVE LOGITS

card

0.35

cards

0.31

boards

0.24

keeper

0.20

keeping

0.20

occo

0.20

CARD

0.19

board

0.19

agli

0.18

keepers

0.18

Activations Density 0.015%

terms related to scoring and evaluations

No Comments

No Known Activations