INDEX

Explanations

dates-based information and product categories

oai_token-act-pair · gpt-3.5-turbo

references to structured data or classifications

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 toget

-0.90

kj

-0.89

 lett

-0.86

 behavi

-0.85

 camer

-0.84

 caus

-0.83

 advoc

-0.83

 eleph

-0.83

 mapping

-0.82

 cabbage

-0.81

POSITIVE LOGITS

 Intermediate

1.54

 Normal

1.41

 Classic

1.40

 Moderate

1.36

 Single

1.36

 Balanced

1.34

 Regular

1.34

 Expert

1.33

 Standard

1.33

 Semi

1.33

Activations Density 0.312%

dates-based information and product categories

references to structured data or classifications

No Comments

No Known Activations