INDEX

Explanations

the repeated use of the word "who," indicating a focus on individuals and their identities

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With LLAMA3.1-8B @ 4-llamascope-res-32k

Configuration

fnlp/Llama3_1-8B-Base-LXR-8x/Llama3_1-8B-Base-L4R-8x

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

cerebras/SlimPajama-627B

Features

32,768

Data Type

bfloat16

Hook Name

blocks.4.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

atic

-0.15

yen

-0.15

tn

-0.15

asn

-0.15

anken

-0.14

kan

-0.14

SURE

-0.13

illes

-0.13

ubi

-0.13

gaard

-0.13

POSITIVE LOGITS

oping

0.20

upon

0.16

OSH

0.15

osh

0.14

hind

0.14

////////////////////////////////////////////////////////////

0.14

akin

0.14

oser

0.13

endale

0.13

’ve

0.13

Activations Density 0.130%

the repeated use of the word "who," indicating a focus on individuals and their identities

No Comments

No Known Activations