INDEX

Explanations

phrases related to duplication, repetition, or copying

oai_token-act-pair · gpt-3.5-turbo

terms associated with replication and mirroring processes

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

anche

-0.90

ember

-0.72

Ty

-0.71

Fir

-0.67

eger

-0.67

 Edgar

-0.67

auts

-0.67

Ãª

-0.66

tera

-0.66

Tar

-0.66

POSITIVE LOGITS

 replicated

2.97

 replication

1.85

 mirrored

1.77

 reproduced

1.19

 Repl

1.09

 perfected

1.03

 copied

0.97

 dupl

0.92

 relational

0.90

 persisted

0.90

Activations Density 0.023%

phrases related to duplication, repetition, or copying

terms associated with replication and mirroring processes

No Comments

No Known Activations