INDEX

Explanations

repeated references to a specific term or name, particularly variations of the letter 'H' followed by numerical designations

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With LLAMA3.1-8B @ 29-llamascope-res-32k

Configuration

fnlp/Llama3_1-8B-Base-LXR-8x/Llama3_1-8B-Base-L29R-8x

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

cerebras/SlimPajama-627B

Features

32,768

Data Type

bfloat16

Hook Name

blocks.29.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

appen

-0.18

ello

-0.18

ackers

-0.17

etros

-0.16

thren

-0.16

idden

-0.16

andin

-0.16

IP

-0.16

á»Ĩ

-0.16

asaki

-0.15

POSITIVE LOGITS

su

0.21

ruby

0.21

rk

0.19

rdf

0.19

sie

0.19

ov

0.18

ulse

0.18

nat

0.18

ocket

0.18

rone

0.17

Activations Density 0.026%

repeated references to a specific term or name, particularly variations of the letter 'H' followed by numerical designations

No Comments

No Known Activations