INDEX

Explanations

text related to sports, specifically football

oai_token-act-pair · gpt-3.5-turbo

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 9-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.9.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.9.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

 Craigslist

-0.75

 Slate

-0.64

 Ballard

-0.63

 Tacoma

-0.62

 Seeking

-0.61

 hypothesized

-0.61

 Supplemental

-0.60

 Spokane

-0.60

Ohio

-0.60

 Lamar

-0.60

POSITIVE LOGITS

 practise

0.78

 learnt

0.77

 afterwards

0.76

 rubbish

0.71

!!

0.70

 civilisation

0.70

`.

0.69

 colours

0.68

 unbelievable

0.68

ana

0.67

Activations Density 1.450%

text related to sports, specifically football

No Comments

No Known Activations