Layer duoc finetune:
[9, 18)
trong tong so
28
layer
(tuong ung khoang 1L/3 -> 2L/3).
Module duoc gan LoRA:
attention
,
router
,
experts
trong khoang layer tren.
r = 16, alpha = 32, dropout = 0.05
Loss
Loss MoE tieu chuan:
L_total = L_LM + lb_loss_coef * L_LB
L_LM
: cross-entropy chuan tren token tiep theo.
L_LB
: load balancing loss chuan cua MoE (Switch/Mixtral style), tinh tren cac router
nam trong khoang layer duoc finetune.
lb_loss_coef
= None
num_experts
= 232,
top_k
= 8
Du lieu
Cau don ngu duoc gom tu 3 bo du lieu alignment:
flores.json
,
bible.json
,
ntrex.json
(moi field ngon ngu trong 1 record duoc coi la 1 sample), shuffle va sort theo do dai token
truoc khi gom batch.
Diagnostics
Xem
diagnostics/loss_log.jsonl
(log theo tung step) va
diagnostics/loss_curve.png
(bieu do L_LM / L_LB / L_Total theo step).
Runs of ducanhdinh Macro-Nano-Instruct-Finetuning on huggingface.co
1.5K
Total runs
-55
24-hour runs
-229
3-day runs
-522
7-day runs
-1.3K
30-day runs
More Information About Macro-Nano-Instruct-Finetuning huggingface.co Model
More Macro-Nano-Instruct-Finetuning license Visit here:
Macro-Nano-Instruct-Finetuning huggingface.co is an AI model on huggingface.co that provides Macro-Nano-Instruct-Finetuning's model effect (), which can be used instantly with this ducanhdinh Macro-Nano-Instruct-Finetuning model. huggingface.co supports a free trial of the Macro-Nano-Instruct-Finetuning model, and also provides paid use of the Macro-Nano-Instruct-Finetuning. Support call Macro-Nano-Instruct-Finetuning model through api, including Node.js, Python, http.
Macro-Nano-Instruct-Finetuning huggingface.co is an online trial and call api platform, which integrates Macro-Nano-Instruct-Finetuning's modeling effects, including api services, and provides a free online trial of Macro-Nano-Instruct-Finetuning, you can try Macro-Nano-Instruct-Finetuning online for free by clicking the link below.
ducanhdinh Macro-Nano-Instruct-Finetuning online free url in huggingface.co:
Macro-Nano-Instruct-Finetuning is an open source model from GitHub that offers a free installation service, and any user can find Macro-Nano-Instruct-Finetuning on GitHub to install. At the same time, huggingface.co provides the effect of Macro-Nano-Instruct-Finetuning install, users can directly use Macro-Nano-Instruct-Finetuning installed effect in huggingface.co for debugging and trial. It also supports api for free installation.
Macro-Nano-Instruct-Finetuning install url in huggingface.co: