INDEX

Explanations

negative evaluations of experiences or products

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

the word "and" when it appears in lists or series of adjectives describing qualities or characteristics.

oai_token-act-pair · claude-4-5-sonnet Triggered by @jyhe0408

the coordinating conjunction used to link paired or listed elements, especially adjoining adjectives.

oai_token-act-pair · gpt-5 Triggered by @jyhe0408

New Auto-Interp

Top Features by Cosine Similarity

Comparing With LLAMA3.1-8B @ 15-llamascope-res-32k

Configuration

fnlp/Llama3_1-8B-Base-LXR-8x/Llama3_1-8B-Base-L15R-8x

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

cerebras/SlimPajama-627B

Features

32,768

Data Type

bfloat16

Hook Name

blocks.15.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

uite

-0.16

 putas

-0.16

uner

-0.15

éļ

-0.15

linkplain

-0.14

_DLL

-0.14

ruba

-0.14

 widest

-0.14

nton

-0.14

-0.13

POSITIVE LOGITS

ajas

0.15

egal

0.14

á»ī

0.14

 -*-č↵

0.14

 duplic

0.14

obi

0.14

/or

0.14

gc

0.14

ãĥ¼ãĥľ

0.14

incinn

0.13

Activations Density 0.184%

negative evaluations of experiences or products

the word "and" when it appears in lists or series of adjectives describing qualities or characteristics.

the coordinating conjunction used to link paired or listed elements, especially adjoining adjectives.

No Comments

No Known Activations

negative evaluations of experiences or products

the word "and" when it appears in lists or series of adjectives describing qualities or characteristics.

the coordinating conjunction used to link paired or listed elements, especially adjoining adjectives.

No Comments

No Known Activations