INDEX

Explanations

terms associated with importance and necessity

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With LLAMA3.1-8B @ 3-llamascope-res-32k

Configuration

fnlp/Llama3_1-8B-Base-LXR-8x/Llama3_1-8B-Base-L3R-8x

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

cerebras/SlimPajama-627B

Features

32,768

Data Type

bfloat16

Hook Name

blocks.3.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

ussian

-0.15

ãĤ¤ãĥ³ãĥĪ

-0.15

oreach

-0.15

_DX

-0.14

rine

-0.14

ilst

-0.14

ocratic

-0.14

OCI

-0.14

repr

-0.14

Ð³Ð¾ÑĢ

-0.14

POSITIVE LOGITS

endir

0.18

meer

0.15

rou

0.15

 Hava

0.15

 importance

0.14

alin

0.14

_effect

0.14

hei

0.14

 Meer

0.14

ãģ°ãģĭãĤĬ

0.14

Activations Density 0.081%

terms associated with importance and necessity

No Comments

No Known Activations