INDEX

Explanations

instances where something is physically attached or fixed onto something else

oai_token-act-pair · gpt-3.5-turbo

instances of the word "mounted."

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

ital

-0.79

Birth

-0.71

uer

-0.68

Gos

-0.67

íķ

-0.66

Dream

-0.66

urity

-0.65

 Gram

-0.64

 Thoughts

-0.64

hyp

-0.63

POSITIVE LOGITS

 mounted

3.76

mounted

2.76

 mounts

2.57

 mount

2.46

 mounting

1.92

mount

1.84

Mount

1.83

 Mount

1.60

 bolted

1.27

 installed

1.19

Activations Density 0.022%

instances where something is physically attached or fixed onto something else

instances of the word "mounted."

No Comments

No Known Activations