INDEX

Explanations

structures that indicate instructional or procedural contexts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With LLAMA3.1-8B @ 11-llamascope-res-32k

Configuration

fnlp/Llama3_1-8B-Base-LXR-8x/Llama3_1-8B-Base-L11R-8x

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

cerebras/SlimPajama-627B

Features

32,768

Data Type

bfloat16

Hook Name

blocks.11.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

edl

-0.16

eniable

-0.15

ptr

-0.15

itol

-0.15

 tane

-0.15

*)((

-0.14

lah

-0.14

_shared

-0.14

anch

-0.14

IMG

-0.14

POSITIVE LOGITS

 forb

0.13

 Punch

0.13

arts

0.13

Amp

0.13

 minus

0.13

pis

0.13

unda

0.12

 loosen

0.12

hol

0.12

Kum

0.12

Activations Density 0.296%

structures that indicate instructional or procedural contexts

No Comments

No Known Activations