INDEX

Explanations

phrases indicating completion or closure

oai_token-act-pair · gpt-3.5-turbo

phrases indicating closure or conclusion

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

ums

-0.75

vere

-0.69

itudes

-0.67

ourge

-0.67

uld

-0.66

outh

-0.64

nuts

-0.63

 masses

-0.63

 Gree

-0.62

bour

-0.62

POSITIVE LOGITS

ãĥ¼ãĥĨ

0.86

âĶĢâĶĢ

0.85

 srfAttach

0.79

actionDate

0.77

REF

0.74

Ñĭ

0.71

 curfew

0.71

jad

0.71

 halt

0.71

 halfway

0.70

Activations Density 0.034%

phrases indicating completion or closure

phrases indicating closure or conclusion

No Comments

No Known Activations