aprender-train 0.31.1

Training & Optimization library with autograd, LoRA, quantization, and model merging
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26

---
source: src/prune/snapshot_tests.rs
assertion_line: 119
expression: config
---
{
  "method": "wanda",
  "target_sparsity": 0.5,
  "pattern": {
    "type": "nm",
    "n": 2,
    "m": 4
  },
  "schedule": {
    "type": "gradual",
    "start_step": 1000,
    "end_step": 5000,
    "initial_sparsity": 0.0,
    "final_sparsity": 0.5,
    "frequency": 100
  },
  "fine_tune_after_pruning": true,
  "fine_tune_steps": 1000,
  "fine_tune_lr": 0.00001,
  "skip_embed_layers": true
}