INDEX

Explanations

prompts asking the reader to take an action

oai_token-act-pair · gpt-3.5-turbo Triggered by @bot

requests for action or error messages related to user interaction

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 12-res-jb

Configuration

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

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_post

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

acebook

-0.64

abwe

-0.64

BIP

-0.63

âĹ¼

-0.61

killed

-0.60

senal

-0.60

ymm

-0.59

emin

-0.59

ukemia

-0.58

fleet

-0.58

POSITIVE LOGITS

try

1.22

 investigate

0.78

Try

0.76

 repeat

0.73

 check

0.73

 hold

0.72

 strive

0.71

 tries

0.71

 attempt

0.71

 tried

0.69

Activations Density 0.004%

prompts asking the reader to take an action

requests for action or error messages related to user interaction

No Comments

No Known Activations