INDEX

Explanations

phrases indicating conjunctions and connections between ideas

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With LLAMA3.1-8B @ 21-llamascope-res-32k

Configuration

fnlp/Llama3_1-8B-Base-LXR-8x/Llama3_1-8B-Base-L21R-8x

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

cerebras/SlimPajama-627B

Features

32,768

Data Type

bfloat16

Hook Name

blocks.21.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

lopedia

-0.16

avig

-0.14

plemented

-0.14

exe

-0.13

iyas

-0.13

ounder

-0.13

ditor

-0.12

cui

-0.12

qui

-0.12

ì¹ĺëĬĶ

-0.12

POSITIVE LOGITS

 that

0.67

 ráº±ng

0.66

that

0.60

 bahwa

0.60

 Å¾e

0.50

	that

0.49

 dass

0.48

 ÑĩÑĤÐ¾

0.47

 Å¼e

0.46

 ÏĮÏĦÎ¹

0.46

Activations Density 0.232%

phrases indicating conjunctions and connections between ideas

No Comments

No Known Activations