INDEX

Explanations

short phrases indicating communication or action

oai_token-act-pair · gpt-3.5-turbo

punctuation and specific short phrases within sentences

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 refinement

-0.77

 flattened

-0.71

 blended

-0.69

 collisions

-0.68

 churn

-0.67

 scattered

-0.65

eren

-0.65

athered

-0.64

 frying

-0.62

 accumulated

-0.62

POSITIVE LOGITS

 Otherwise

0.93

 ASAP

0.90

 Currently

0.84

 Would

0.82

ļéĨĴ

0.81

 Specifically

0.79

 Allows

0.76

 Please

0.76

 Ideally

0.76

Otherwise

0.75

Activations Density 0.712%

short phrases indicating communication or action

punctuation and specific short phrases within sentences

No Comments

No Known Activations