INDEX

Explanations

words related to positive statements or affirmations

oai_token-act-pair · gpt-3.5-turbo

expressions of mixed emotions or reflections on experiences

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

 OECD

-0.57

 indigenous

-0.53

zu

-0.53

 selectively

-0.52

 predec

-0.52

 Include

-0.52

 indo

-0.52

Similarly

-0.52

 prolifer

-0.52

 quotas

-0.51

POSITIVE LOGITS

 disappointment

1.12

 congratulations

1.08

 disappointing

1.01

 sadness

1.00

 congr

0.97

 Congratulations

0.93

wow

0.93

 emotions

0.90

 disappointed

0.90

 saddened

0.90

Activations Density 0.533%

words related to positive statements or affirmations

expressions of mixed emotions or reflections on experiences

No Comments

No Known Activations