INDEX

Explanations

responses or replies

oai_token-act-pair · gpt-3.5-turbo

instances of dialogue or responses

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

 historic

-0.62

 torch

-0.61

 elev

-0.57

 horrific

-0.57

 consolidation

-0.56

 film

-0.54

 intra

-0.53

 microscope

-0.53

 impair

-0.53

 hard

-0.53

POSITIVE LOGITS

 replied

3.37

 answered

2.46

 responded

2.41

 replies

2.36

 reply

2.18

 responds

1.86

 countered

1.60

answered

1.59

 nodded

1.56

 exclaimed

1.55

Activations Density 0.021%

responses or replies

instances of dialogue or responses

No Comments

No Known Activations