INDEX

Explanations

words related to fire

oai_token-act-pair · gpt-3.5-turbo

terms related to fire and related incidents

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

ļéĨĴ

-0.68

BST

-0.64

entric

-0.62

 abstract

-0.61

 capsules

-0.61

 plag

-0.60

Ô

-0.60

cm

-0.60

 McGill

-0.59

 discontin

-0.58

POSITIVE LOGITS

fire

4.53

fires

3.65

Fire

2.74

 fire

2.13

fired

2.13

flame

2.01

 Fire

1.94

 FIRE

1.74

 fires

1.71

 Fires

1.47

Activations Density 0.008%

words related to fire

terms related to fire and related incidents

No Comments

No Known Activations