INDEX

Explanations

positive descriptions and evaluations

oai_token-act-pair · gpt-3.5-turbo

expressions of subjective opinion or evaluation

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

ļéĨĴ

-0.77

failed

-0.66

ãĤ´ãĥ³

-0.66

 Alone

-0.65

ourning

-0.64

çĶ°

-0.63

ifled

-0.63

den

-0.63

but

-0.62

ãĥĨ

-0.62

POSITIVE LOGITS

 nonetheless

1.55

 nevertheless

1.24

 worth

1.22

 worthwhile

1.19

 undeniable

1.16

 undeniably

1.13

 manageable

1.09

 achievable

1.00

 damn

1.00

 definitely

0.97

Activations Density 0.398%

positive descriptions and evaluations

expressions of subjective opinion or evaluation

No Comments

No Known Activations