INDEX

Explanations

words related to comparisons where one value is smaller than another

oai_token-act-pair · gpt-3.5-turbo

references to lower quantities or reduced values

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

own

-0.68

ynasty

-0.65

etz

-0.65

eta

-0.64

esh

-0.62

anka

-0.61

este

-0.61

chuk

-0.60

ortium

-0.58

anon

-0.58

POSITIVE LOGITS

 less

3.06

 fewer

2.38

 Less

1.77

 lesser

1.73

Less

1.68

 weaker

1.62

 slower

1.60

 shorter

1.57

 more

1.55

 poorer

1.46

Activations Density 0.029%

words related to comparisons where one value is smaller than another

references to lower quantities or reduced values

No Comments

No Known Activations