INDEX

Explanations

phrases or statements where something is being perceived, considered, or pondered

oai_token-act-pair · gpt-3.5-turbo

instances of the word "thought" 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

ember

-0.70

 attm

-0.68

achable

-0.64

 RELEASE

-0.63

rx

-0.62

plane

-0.61

halla

-0.59

mate

-0.59

Prov

-0.59

 Provide

-0.59

POSITIVE LOGITS

 thought

3.56

 Thought

2.42

thought

2.28

 believed

1.75

 thinking

1.58

 imagined

1.58

 thoughts

1.49

 reasoned

1.44

 thinks

1.34

 think

1.34

Activations Density 0.026%

phrases or statements where something is being perceived, considered, or pondered

instances of the word "thought" in various contexts

No Comments

No Known Activations