INDEX

Explanations

phrases related to technology, possibly technical errors or issues

oai_token-act-pair · gpt-3.5-turbo

symbols and markers indicating emphasis or special notation

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

 guarding

-0.70

swe

-0.68

 tyres

-0.62

 manif

-0.62

 union

-0.60

doors

-0.60

agall

-0.59

 Cull

-0.58

wagen

-0.58

 travers

-0.57

POSITIVE LOGITS

mosp

0.87

maxwell

0.79

 Hilbert

0.76

 References

0.75

 Comments

0.69

ategor

0.68

HAM

0.67

-+-+-+-+

0.66

Â¯Â¯Â¯Â¯Â¯Â¯Â¯Â¯Â¯Â¯Â¯Â¯Â¯Â¯Â¯Â¯

0.66

idav

0.66

Activations Density 0.242%

phrases related to technology, possibly technical errors or issues

symbols and markers indicating emphasis or special notation

No Comments

No Known Activations