INDEX

Explanations

phrases related to sequels or follow-up actions

oai_token-act-pair · gpt-3.5-turbo

phrases related to follow-up actions or reports

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

 nerv

-0.70

 DRAG

-0.68

 Burr

-0.63

 Graveyard

-0.63

 Cavern

-0.62

bage

-0.62

 Powder

-0.61

 Shades

-0.60

 Nicotine

-0.60

 Smoke

-0.59

POSITIVE LOGITS

sized

0.93

credit

0.93

advertisement

0.88

sent

0.87

only

0.87

minded

0.86

through

0.86

turned

0.85

feature

0.85

heavy

0.85

Activations Density 0.088%

phrases related to sequels or follow-up actions

phrases related to follow-up actions or reports

No Comments

No Known Activations