INDEX

Explanations

phrases containing the word "haven" that suggest hesitation or lack of completion or engagement

oai_token-act-pair · gpt-3.5-turbo

instances of the word "haven."

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

forms

-0.69

bert

-0.66

 Article

-0.61

Ext

-0.61

als

-0.61

tones

-0.60

ulz

-0.60

intent

-0.58

Fact

-0.58

grades

-0.57

POSITIVE LOGITS

 haven

3.44

haven

2.08

 hasn

1.87

 hadn

1.85

 aren

1.64

 Haven

1.61

 weren

1.56

 havens

1.50

don

1.28

 didn

1.23

Activations Density 0.010%

phrases containing the word "haven" that suggest hesitation or lack of completion or engagement

instances of the word "haven."

No Comments

No Known Activations