INDEX

Explanations

references to financial trading or transactions

oai_token-act-pair · gpt-3.5-turbo

occurrences of the word "traded."

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

 Alam

-0.70

 slideshow

-0.68

 Insp

-0.65

RECT

-0.62

 Jennifer

-0.61

Dul

-0.60

PS

-0.60

document

-0.59

 Lisa

-0.59

cule

-0.58

POSITIVE LOGITS

 traded

3.74

 trading

1.95

 exchanged

1.81

 trades

1.80

trade

1.77

Trade

1.77

 trade

1.73

 Trade

1.54

 Trading

1.52

 swapped

1.51

Activations Density 0.010%

references to financial trading or transactions

occurrences of the word "traded."

No Comments

No Known Activations