INDEX

Explanations

instances of the prefix "pro-" and related variations

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With LLAMA3.1-8B @ 26-llamascope-res-32k

Configuration

fnlp/Llama3_1-8B-Base-LXR-8x/Llama3_1-8B-Base-L26R-8x

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

cerebras/SlimPajama-627B

Features

32,768

Data Type

bfloat16

Hook Name

blocks.26.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

Pro

-0.20

 Pros

-0.18

Pro

-0.18

-0.17

nic

-0.16

ags

-0.16

cxx

-0.16

Ã³

-0.15

pent

-0.15

WithMany

-0.15

POSITIVE LOGITS

vision

0.20

pery

0.19

tracted

0.19

portion

0.17

pped

0.16

owler

0.16

trait

0.16

spect

0.16

verbs

0.16

cura

0.16

Activations Density 0.044%

instances of the prefix "pro-" and related variations

No Comments

No Known Activations