INDEX

Explanations

terms related to indirectness or secondary effects

oai_token-act-pair · gpt-3.5-turbo

references to indirect relationships or connections

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

ï¸ı

-0.81

suits

-0.80

urus

-0.71

apers

-0.71

Enjoy

-0.70

 stormed

-0.69

suit

-0.69

 Featuring

-0.69

Ready

-0.68

rock

-0.68

POSITIVE LOGITS

 indirect

2.47

 implicit

2.01

 conditional

1.94

 incremental

1.77

 implied

1.61

 indirectly

1.58

 circumst

1.54

 incidental

1.51

 hypothetical

1.49

 differential

1.48

Activations Density 0.056%

terms related to indirectness or secondary effects

references to indirect relationships or connections

No Comments

No Known Activations