INDEX

Explanations

terms specifying particular categories or types of items

oai_token-act-pair · gpt-3.5-turbo

references to specifications or specific details

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

ga

-0.67

esc

-0.63

 behind

-0.63

 mortgage

-0.62

 pile

-0.62

 swept

-0.62

bos

-0.62

 indef

-0.61

 scam

-0.61

bet

-0.60

POSITIVE LOGITS

specific

3.73

Specific

2.20

cific

1.71

 Specific

1.62

 specific

1.60

related

1.58

 specificity

1.57

relevant

1.45

appropriate

1.35

exclusive

1.30

Activations Density 0.013%

terms specifying particular categories or types of items

references to specifications or specific details

No Comments

No Known Activations