INDEX

Explanations

terms related to detailed scrutiny or examination

oai_token-act-pair · gpt-3.5-turbo

references to various types of analyses

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

ovable

-0.82

Ń·

-0.69

psc

-0.68

ummer

-0.67

oppable

-0.66

pire

-0.65

²¾

-0.65

Park

-0.64

resses

-0.63

iah

-0.63

POSITIVE LOGITS

 analysis

3.50

 analyses

2.75

 Analysis

2.58

analysis

2.29

Analysis

2.07

 analys

1.92

 analyzing

1.85

alyses

1.85

 assessment

1.83

 analyze

1.81

Activations Density 0.027%

terms related to detailed scrutiny or examination

references to various types of analyses

No Comments

No Known Activations