INDEX

Explanations

terms related to presentation or proposals

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With LLAMA3.1-8B @ 23-llamascope-res-32k

Configuration

fnlp/Llama3_1-8B-Base-LXR-8x/Llama3_1-8B-Base-L23R-8x

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

cerebras/SlimPajama-627B

Features

32,768

Data Type

bfloat16

Hook Name

blocks.23.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

Ìĥ

-0.18

 flushed

-0.16

vit

-0.15

ear

-0.15

 flushing

-0.15

/mit

-0.15

amba

-0.14

 flush

-0.14

viz

-0.14

isch

-0.14

POSITIVE LOGITS

fork

0.49

 fork

0.23

pitch

0.21

 Fork

0.20

 pitch

0.20

 forks

0.20

(es

0.20

 Pitch

0.19

 pitched

0.18

cock

0.18

Activations Density 0.007%

terms related to presentation or proposals

No Comments

No Known Activations