INDEX

Explanations

dates mentioned in a specific format

oai_token-act-pair · gpt-3.5-turbo

dates and significant temporal references

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

 suite

-0.70

 abbre

-0.66

=-=-=-=-=-=-=-=-

-0.63

 persecuted

-0.61

 chang

-0.59

=-=-=-=-

-0.57

ersed

-0.57

 DRAG

-0.56

free

-0.56

 norm

-0.56

POSITIVE LOGITS

th

0.80

 occasions

0.72

Thom

0.70

rum

0.68

 Thom

0.67

 Kass

0.66

mid

0.65

elt

0.64

ushima

0.64

lem

0.62

Activations Density 0.153%

dates mentioned in a specific format

dates and significant temporal references

No Comments

No Known Activations