INDEX

Explanations

instances of the prefix "re-" indicating repetition or restoration actions

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With LLAMA3.1-8B @ 30-llamascope-res-32k

Configuration

fnlp/Llama3_1-8B-Base-LXR-8x/Llama3_1-8B-Base-L30R-8x

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

cerebras/SlimPajama-627B

Features

32,768

Data Type

bfloat16

Hook Name

blocks.30.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

olo

-0.16

íİ¸

-0.15

intl

-0.15

 retro

-0.14

.ToShort

-0.14

GES

-0.14

rey

-0.14

mÃ©na

-0.13

andes

-0.13

lum

-0.13

POSITIVE LOGITS

search

0.22

known

0.18

earch

0.17

ward

0.17

istrar

0.17

pson

0.16

ponsive

0.16

 search

0.16

ource

0.16

ason

0.16

Activations Density 0.028%

instances of the prefix "re-" indicating repetition or restoration actions

No Comments

No Known Activations