INDEX

Explanations

phrases related to direction or guidance

oai_token-act-pair · gpt-3.5-turbo

assertions about directionality, particularly regarding positive or negative trajectories

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

Sec

-0.62

ulas

-0.61

 Splash

-0.60

çīĪ

-0.60

Availability

-0.58

arcity

-0.58

 fame

-0.58

yrs

-0.57

 Instit

-0.56

Spa

-0.56

POSITIVE LOGITS

 direction

2.09

 directions

1.83

 footsteps

1.31

direction

1.22

 Direction

1.18

 opposite

1.11

wards

0.99

 favor

0.95

WARD

0.93

ward

0.92

Activations Density 0.148%

phrases related to direction or guidance

assertions about directionality, particularly regarding positive or negative trajectories

No Comments

No Known Activations