INDEX

Explanations

words related to politics and specific entities for whom the neuron has a favorable disposition

oai_token-act-pair · gpt-3.5-turbo

phrases related to significant societal themes or critiques

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.hook_resid_pre

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

torch.float32

Hook Point

blocks.10.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 doesnt

-0.54

 "$:/

-0.52

DragonMagazine

-0.51

 didnt

-0.49

 wont

-0.48

 learnt

-0.45

tyr

-0.44

)",

-0.44

 alot

-0.44

©¶æ

-0.44

POSITIVE LOGITS

—

0.93

0.91

—

0.88

—"

0.87

 âĢķ

0.82

.;

0.78

--

0.72

.*

0.72

0.71

Activations Density 1.330%

words related to politics and specific entities for whom the neuron has a favorable disposition

phrases related to significant societal themes or critiques

No Comments

No Known Activations