INDEX

Explanations

phrases indicating steps or instructions

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With LLAMA3.1-8B @ 25-llamascope-res-32k

Configuration

fnlp/Llama3_1-8B-Base-LXR-8x/Llama3_1-8B-Base-L25R-8x

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

cerebras/SlimPajama-627B

Features

32,768

Data Type

bfloat16

Hook Name

blocks.25.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

avou

-0.17

stor

-0.16

åŁº

-0.15

heimer

-0.15

roe

-0.15

istik

-0.14

ê´Ģ

-0.14

 decis

-0.14

ãĤ¤ãĥ³ãĥĪ

-0.14

osas

-0.14

POSITIVE LOGITS

sie

0.17

ei

0.16

 Schul

0.15

eat

0.15

.sell

0.15

fly

0.14

fly

0.14

 sell

0.14

 Schultz

0.14

 Appro

0.14

Activations Density 0.158%

phrases indicating steps or instructions

No Comments

No Known Activations