INDEX

Explanations

references to the concept of being "in" or indicating location and involvement

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

Looking at the activations, this neuron activates most strongly on the phrase "in" followed by words or contexts that suggest danger, risk, trouble, or precarious situations (such as "jeopardy", "serious", "balance", "full swing", "the mix", "

oai_token-act-pair · claude-4-5-sonnet Triggered by @jyhe0408

New Auto-Interp

Top Features by Cosine Similarity

Comparing With LLAMA3.1-8B @ 15-llamascope-res-32k

Configuration

fnlp/Llama3_1-8B-Base-LXR-8x/Llama3_1-8B-Base-L15R-8x

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

cerebras/SlimPajama-627B

Features

32,768

Data Type

bfloat16

Hook Name

blocks.15.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

ugins

-0.14

Ø±Ùħ

-0.14

edu

-0.14

eken

-0.14

anik

-0.14

fq

-0.14

à¸ĩà¸²à¸Ļ

-0.14

_SM

-0.14

edb

-0.14

ç´Ģ

-0.14

POSITIVE LOGITS

oreal

0.18

 graf

0.14

agus

0.14

ilet

0.14

 Ø¯Ø±ÛĮ

0.14

map

0.14

XHR

0.13

uli

0.13

 GRAPH

0.13

PLICATION

0.13

Activations Density 0.082%

references to the concept of being "in" or indicating location and involvement

Looking at the activations, this neuron activates most strongly on the phrase "in" followed by words or contexts that suggest danger, risk, trouble, or precarious situations (such as "jeopardy", "serious", "balance", "full swing", "the mix", "

No Comments

No Known Activations

references to the concept of being "in" or indicating location and involvement

Looking at the activations, this neuron activates most strongly on the phrase "in" followed by words or contexts that suggest danger, risk, trouble, or precarious situations (such as "jeopardy", "serious", "balance", "full swing", "the mix", "

No Comments

No Known Activations