INDEX

Explanations

references to "behind the scenes" or similar expressions indicating unseen processes or activities

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

osten

-0.16

.scalablytyped

-0.15

dit

-0.15

OLS

-0.15

ongan

-0.14

uv

-0.14

igate

-0.14

èĬ³

-0.14

 deferred

-0.13

ULONG

-0.13

POSITIVE LOGITS

/background

0.15

Luk

0.15

opus

0.15

warm

0.15

 warm

0.15

ku

0.14

 kole

0.14

/commons

0.14

 Warm

0.14

 warmed

0.14

Activations Density 0.007%

references to "behind the scenes" or similar expressions indicating unseen processes or activities

No Comments

No Known Activations