INDEX

Explanations

terms associated with placeholders and vague references

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With LLAMA3.1-8B @ 27-llamascope-res-32k

Configuration

fnlp/Llama3_1-8B-Base-LXR-8x/Llama3_1-8B-Base-L27R-8x

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

cerebras/SlimPajama-627B

Features

32,768

Data Type

bfloat16

Hook Name

blocks.27.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

ersh

-0.16

wert

-0.15

createCommand

-0.15

FieldValue

-0.15

.Fields

-0.14

 field

-0.14

 Hurt

-0.14

 åł

-0.14

GetValue

-0.14

onga

-0.14

POSITIVE LOGITS

 sight

0.17

 coverage

0.15

ason

0.15

onda

0.15

azz

0.15

 CLUB

0.15

rust

0.14

trand

0.14

coverage

0.14

 STORAGE

0.14

Activations Density 0.019%

terms associated with placeholders and vague references

No Comments

No Known Activations