Decision Workspace
float4 vs entrenar vs torsh-quantization
Side-by-side comparison of Rust crates
60
float4
experimentalv0.2.0
MXFP4-compatible 4-bit floating point types and block formats for Rust.
60
entrenar
experimentalv0.7.6
Training & Optimization library with autograd, LoRA, quantization, and model merging
59
torsh-quantization
experimentalv0.1.1
Model quantization for ToRSh neural networks
Core Metrics
| float4 | entrenar | torsh-quantization | |
|---|---|---|---|
| Health Score | 60 | 60 | 59 |
| Total Downloads | 204.0K | 11.9K | 563 |
| 30d Downloads | 52.0K | 1.3K | 62 |
| Dependents | 29 | 119 | 7 |
| Releases | 2 | 30 | 6 |
| Last Updated | 23d ago | 13d ago | 7d ago |
| Age | 7m | 4m | 5m |
Health Breakdown
float4
Maintenance
16
Quality
15
Community
10
Popularity
6
Documentation
13
entrenar
Maintenance
19
Quality
12
Community
12
Popularity
5
Documentation
12
torsh-quantization
Maintenance
20
Quality
13
Community
8
Popularity
3
Documentation
15
Technical Details
| float4 | entrenar | torsh-quantization | |
|---|---|---|---|
| Version | 0.2.0 | 0.7.6 | 0.1.1 |
| Stable (≥1.0) | ✗ No | ✗ No | ✗ No |
| License | MIT | MIT | Apache-2.0 |
| Dependencies | 0 | 61 | 11 |
| Crate Size | 17KB | 1.5MB | 277KB |
| Features | 0 | 17 | 3 |
| Yanked % | 0.0% | 0.0% | 0.0% |
| Edition | 2024 | 2021 | 2021 |
| MSRV | — | 1.87 | 1.77 |
| Owners | 1 | 1 | 1 |
Links
Quick Verdict
- •float4 leads with a health score of 60/100, but none of the options score above 80.
- •entrenar is depended on by 119 crates — strongest ecosystem trust.