INDEX

Explanations

instances of the word "approve" in different contexts

oai_token-act-pair · gpt-3.5-turbo

instances of the word "approve" and its variations 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

 Falling

-0.61

 Galactic

-0.61

 leve

-0.61

Kag

-0.60

 Desmond

-0.59

izz

-0.59

 Shank

-0.59

stack

-0.58

di

-0.58

Hem

-0.58

POSITIVE LOGITS

 approve

3.73

 approves

2.60

 disapprove

2.39

 approving

2.23

 approval

1.99

 approved

1.96

 authorize

1.85

 approvals

1.80

 Appro

1.64

 disappro

1.64

Activations Density 0.012%

instances of the word "approve" in different contexts

instances of the word "approve" and its variations in various contexts

No Comments

No Known Activations