INDEX

Explanations

phrases related to product features and classifications

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With LLAMA3.1-8B @ 14-llamascope-res-32k

Configuration

fnlp/Llama3_1-8B-Base-LXR-8x/Llama3_1-8B-Base-L14R-8x

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

cerebras/SlimPajama-627B

Features

32,768

Data Type

bfloat16

Hook Name

blocks.14.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

 griev

-0.16

bach

-0.15

ikan

-0.15

phia

-0.15

acht

-0.14

 dred

-0.14

 noqa

-0.14

achuset

-0.13

cox

-0.13

licken

-0.13

POSITIVE LOGITS

 direct

0.23

 instead

0.21

 directly

0.21

çĽ´æİ¥

0.21

direct

0.19

 Instead

0.18

Instead

0.18

instead

0.18

 Direct

0.17

oen

0.17

Activations Density 0.214%

phrases related to product features and classifications

No Comments

No Known Activations