INDEX

Explanations

words related to balancing or trading off actions or decisions

oai_token-act-pair · gpt-3.5-turbo

references to trade-offs and their consequences

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 existence

-0.59

 update

-0.59

 Story

-0.56

 Nicotine

-0.55

 stat

-0.55

 stamp

-0.55

 Cricket

-0.54

 nicotine

-0.54

 Sheila

-0.54

 breathing

-0.53

POSITIVE LOGITS

offs

4.51

off

2.59

OFF

1.85

downs

1.55

Off

1.51

outs

1.46

aways

1.44

ups

1.38

overs

1.23

backs

1.22

Activations Density 0.008%

words related to balancing or trading off actions or decisions

references to trade-offs and their consequences

No Comments

No Known Activations