INDEX

Explanations

words related to actions, particularly those ending in 'ing' or 'er'

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

New Auto-Interp

Top Features by Cosine Similarity

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

Configuration

fnlp/Llama3_1-8B-Base-LXR-8x/Llama3_1-8B-Base-L1R-8x

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

cerebras/SlimPajama-627B

Features

32,768

Data Type

bfloat16

Hook Name

blocks.1.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

oley

-0.15

.setAction

-0.15

undler

-0.14

quam

-0.14

resco

-0.14

esson

-0.14

onse

-0.14

snap

-0.14

essor

-0.14

otoxic

-0.14

POSITIVE LOGITS

ered

0.15

/archive

0.14

eri

0.14

 Seal

0.14

 Baldwin

0.14

laden

0.13

Bai

0.13

iddi

0.13

adeon

0.13

NAL

0.13

Activations Density 0.039%

words related to actions, particularly those ending in 'ing' or 'er'

No Comments

No Known Activations