INDEX

Explanations

dates represented in specific formats

oai_token-act-pair · gpt-3.5-turbo

sequences of numbers or dates

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

 scrap

-0.73

ername

-0.71

ĸļ

-0.70

ĺħ

-0.69

 governed

-0.67

 mathemat

-0.65

 administr

-0.65

wat

-0.64

 arri

-0.62

 altogether

-0.62

POSITIVE LOGITS

âĵĺ

0.87

________________________________________________________________

0.85

Conclusion

0.80

Frames

0.79

Else

0.78

--------------------

0.77

Sem

0.73

else

0.73

Chest

0.73

Instruct

0.72

Activations Density 0.236%

dates represented in specific formats

sequences of numbers or dates

No Comments

No Known Activations