INDEX

Explanations

the word "a"

oai_token-act-pair · gpt-3.5-turbo

frequency of articles and adjectives in reference to quantities or counts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

hyde

-0.63

information

-0.62

 endeavour

-0.59

iliation

-0.59

matter

-0.58

 activism

-0.58

Finish

-0.57

 Advice

-0.56

 expertise

-0.56

peed

-0.55

POSITIVE LOGITS

few

1.44

 handful

1.32

 couple

1.28

 dozen

1.22

 bunch

1.17

lot

1.17

dozen

1.12

few

1.12

 plethora

1.10

 number

1.10

Activations Density 0.060%

the word "a"

frequency of articles and adjectives in reference to quantities or counts

No Comments

No Known Activations