INDEX

Explanations

words related to fine-tuning or adjustments

oai_token-act-pair · gpt-3.5-turbo

parts of words or suffixes associated with word formation

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

 hardened

-0.64

SPONSORED

-0.63

izoph

-0.62

otropic

-0.60

herry

-0.60

 representations

-0.60

 sheer

-0.60

 embodiments

-0.59

 thick

-0.59

 scripture

-0.59

POSITIVE LOGITS

camp

0.82

 Kemp

0.75

atis

0.71

ports

0.69

 Davis

0.66

 Daniels

0.63

 Tags

0.61

Camp

0.60

sburgh

0.60

Davis

0.60

Activations Density 0.320%

words related to fine-tuning or adjustments

parts of words or suffixes associated with word formation

No Comments

No Known Activations