INDEX

Explanations

words related to final outcomes or conclusions

oai_token-act-pair · gpt-3.5-turbo

expressions indicating final outcomes or conclusions

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

 greeting

-0.65

hift

-0.63

 Prayer

-0.62

pop

-0.62

 Bulgar

-0.59

:"

-0.56

 google

-0.56

load

-0.56

 dozen

-0.55

 Regular

-0.54

POSITIVE LOGITS

 ultimately

3.34

 eventually

2.47

Ultimately

2.04

 Ultimately

2.01

 inevitably

1.64

 Eventually

1.50

Eventually

1.50

 eventual

1.49

 subsequently

1.45

 finally

1.42

Activations Density 0.018%

words related to final outcomes or conclusions

expressions indicating final outcomes or conclusions

No Comments

No Known Activations