INDEX

Explanations

personal opinions and evaluations in a text

oai_token-act-pair · gpt-3.5-turbo

personal admissions and reflections

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

 srfAttach

-0.80

 coerc

-0.74

 Shutterstock

-0.73

 disobedience

-0.73

 lawful

-0.73

 subsistence

-0.72

 unlawful

-0.71

 liberation

-0.69

 extraction

-0.69

 asylum

-0.69

POSITIVE LOGITS

 Prediction

1.28

 predicting

1.13

ichick

1.10

 pegged

1.08

 prediction

1.06

 predict

1.04

ESPN

1.02

 scouts

1.02

Pred

1.02

 pundits

1.00

Activations Density 0.424%

personal opinions and evaluations in a text

personal admissions and reflections

No Comments

No Known Activations