INDEX

Explanations

phrases related to mismatch or non-alignment

oai_token-act-pair · gpt-3.5-turbo

terms related to fitting or compatibility

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

 Moonlight

-0.71

pap

-0.66

 leaves

-0.65

 snail

-0.63

 concentrations

-0.63

 molecules

-0.62

 contents

-0.62

 verbal

-0.61

ch

-0.60

 Secret

-0.59

POSITIVE LOGITS

fit

4.78

fits

3.71

fitting

3.31

fitted

2.53

Fit

2.32

Fit

1.59

fit

1.31

suits

1.21

fell

1.19

 fitt

1.18

Activations Density 0.014%

phrases related to mismatch or non-alignment

terms related to fitting or compatibility

No Comments

No Known Activations