INDEX

Explanations

words related to being against something or having a negative connotation

oai_token-act-pair · gpt-3.5-turbo

terms related to prohibition or restrictive measures

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

ail

-0.67

 anchor

-0.65

 airst

-0.64

 Shades

-0.63

 bottled

-0.63

 coping

-0.62

 enjoyed

-0.61

shr

-0.61

Ic

-0.60

 slashed

-0.60

POSITIVE LOGITS

pro

3.87

Pro

2.17

PRO

1.62

Pro

1.61

prop

1.46

PRO

1.43

prot

1.38

pro

1.29

prof

1.19

pr

1.17

Activations Density 0.012%

words related to being against something or having a negative connotation

terms related to prohibition or restrictive measures

No Comments

No Known Activations