INDEX

Explanations

all forms of the word "to" and its variations

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

Ã¶h

-0.17

ogram

-0.16

jom

-0.15

compat

-0.15

xba

-0.15

andon

-0.14

uzu

-0.14

Compat

-0.14

izo

-0.14

POSITIVE LOGITS

 admit

0.32

 admitting

0.25

 confess

0.24

 admission

0.23

 admits

0.23

 confession

0.22

 admitted

0.22

 honest

0.20

 admissions

0.20

 confessed

0.19

Activations Density 0.028%

all forms of the word "to" and its variations

No Comments

No Known Activations