INDEX

Explanations

numerical sequences or series following a specific pattern, typically involving counting upwards or in steps

oai_token-act-pair · gpt-3.5-turbo

mentions of sequential parts or sections in a structured document

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

atan

-0.72

 passer

-0.70

etimes

-0.68

athered

-0.66

ually

-0.64

ordinate

-0.62

lore

-0.62

ocial

-0.61

ustomed

-0.60

amina

-0.60

POSITIVE LOGITS

 onwards

0.92

 onward

0.79

Dialogue

0.63

 Hundred

0.62

ħĭ

0.60

 Conclusion

0.60

ç«

0.60

 Recap

0.59

 Execution

0.59

Activations Density 0.095%

numerical sequences or series following a specific pattern, typically involving counting upwards or in steps

mentions of sequential parts or sections in a structured document

No Comments

No Known Activations