INDEX

Explanations

numerical values

oai_token-act-pair · gpt-3.5-turbo

references to aggregate quantities or totals in various contexts

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

think

-0.73

asso

-0.69

inks

-0.69

\<

-0.68

unks

-0.63

 Grounds

-0.62

lake

-0.62

ink

-0.61

onson

-0.60

iddler

-0.60

POSITIVE LOGITS

 total

3.63

total

2.68

 TOTAL

2.42

Total

2.15

 Total

2.04

 totals

1.79

 maximum

1.55

 overall

1.48

 complete

1.41

OTAL

1.41

Activations Density 0.014%

numerical values

references to aggregate quantities or totals in various contexts

No Comments

No Known Activations