INDEX

Explanations

words related to ratings, such as "approval rating" and "credit rating."

oai_token-act-pair · gpt-3.5-turbo

terms related to various types of ratings, specifically credit and approval ratings

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

aug

-0.74

vas

-0.65

 Engels

-0.65

achus

-0.63

rag

-0.63

ulpt

-0.63

esta

-0.63

aunder

-0.62

bians

-0.62

ket

-0.62

POSITIVE LOGITS

 rating

3.90

 Rating

3.01

 ratings

2.77

Rating

2.53

 Ratings

2.27

 rated

2.10

rating

2.02

 ranking

1.57

 Rated

1.55

rated

1.39

Activations Density 0.018%

words related to ratings, such as "approval rating" and "credit rating."

terms related to various types of ratings, specifically credit and approval ratings

No Comments

No Known Activations