INDEX

Explanations

locations or positions within a physical space, especially mentioning high or upper regions

oai_token-act-pair · gpt-3.5-turbo

the repeated mention of "upper" related to various contexts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 9-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.9.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.9.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

eln

-0.77

printf

-0.75

tnc

-0.74

 Machina

-0.73

inatory

-0.72

 Chaser

-0.69

atile

-0.69

ensable

-0.68

MRI

-0.66

POSITIVE LOGITS

most

1.31

 bounds

0.89

 tiers

0.84

 limb

0.84

class

0.83

 lobe

0.82

 reaches

0.79

lip

0.78

 extrem

0.78

 tier

0.78

Activations Density 0.031%

locations or positions within a physical space, especially mentioning high or upper regions

the repeated mention of "upper" related to various contexts

No Comments

No Known Activations