INDEX

Explanations

Conjunctions / prepositions

np_max-act · gemini-2.0-flash

This neuron doesn’t pick out any coherent semantic category; it simply gives low‐level bumps on common function words and punctuation—essentially behaving like a noisy “filler” detector rather than looking for any particular content.

oai_token-act-pair · o4-mini Triggered by @xinyanhu8

New Auto-Interp

Configuration

andyrdt/saes-llama-3.1-8b-instruct/resid_post_layer_11/trainer_1

Dataset (Dashboard)

Various

Features

131,072

Data Type

float32

Hook Name

blocks.11.hook_resid_post

Architecture

standard

Context Size

1,024

Dataset

monology/pile-uncopyrighted

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 gentlemen

-0.07

weeks

-0.07

<location

-0.07

 countless

-0.06

 supermarket

-0.06

 mList

-0.06

.Enum

-0.06

(template

-0.06

 Manufact

-0.06

dma

-0.06

POSITIVE LOGITS

�

0.07

レ

0.06

ستم

0.06

()">↵

0.06

ogy

0.06

nb

0.06

 nombres

0.06

 уход

0.06

 واقعی

0.06

凌

0.06

Activations Density 0.027%

Conjunctions / prepositions

This neuron doesn’t pick out any coherent semantic category; it simply gives low‐level bumps on common function words and punctuation—essentially behaving like a noisy “filler” detector rather than looking for any particular content.

No Comments

No Known Activations

Conjunctions / prepositions

This neuron doesn’t pick out any coherent semantic category; it simply gives low‐level bumps on common function words and punctuation—essentially behaving like a noisy “filler” detector rather than looking for any particular content.

No Comments

No Known Activations