mradermacher/L3-Umbral-Mind-RP-v0.3-8B-i1-GGUF
Updated
•
16
•
2
Mendonca
Dihelson
AI & ML interests
None yet
Recent Activity
liked
a model
3 days ago
mradermacher/L3-Umbral-Mind-RP-v0.3-8B-i1-GGUF
liked
a model
3 days ago
mradermacher/Reasoning-Llama-3.1-CoT-RE1-NMT-V2-ORPO-i1-GGUF
liked
a model
4 days ago
Casual-Autopsy/L3-Super-Nova-RP-8B-4.3bpw-h6-exl2
Organizations
None yet
Dihelson's activity
liked
2
models
3 days ago
liked
3
models
4 days ago
liked
a
model
13 days ago
liked
a
Space
6 months ago
reacted to
Jaward's
post with 👀❤️
6 months ago
Post
1786
Supercool Weekend Read🤖
Nvidia researchers achieved SOTA LLM compression metrics using pruning and knowledge distillation techniques.
Details on Techniques (Simplified):
They started off with a large pre-trained language model (15B params), then:
1. Estimated the importance of different parts of the model (neurons, attention heads, layers) using activation-based metrics on a small calibration dataset.
2. Pruned (remove) less important parts of the model to reduce its size.
3. Retrained the pruned model using knowledge distillation, where the original large model acts as a teacher for the smaller pruned model.
4. Used a lightweight neural architecture search to find the best configuration for the pruned model.
5. Repeated this process iteratively to create even smaller models.
Cool, giving it a try this weekend 😎
Code: https://github.com/NVlabs/Minitron
Paper: https://arxiv.org/abs/2407.14679
Demo: nvidia/minitron
Nvidia researchers achieved SOTA LLM compression metrics using pruning and knowledge distillation techniques.
Details on Techniques (Simplified):
They started off with a large pre-trained language model (15B params), then:
1. Estimated the importance of different parts of the model (neurons, attention heads, layers) using activation-based metrics on a small calibration dataset.
2. Pruned (remove) less important parts of the model to reduce its size.
3. Retrained the pruned model using knowledge distillation, where the original large model acts as a teacher for the smaller pruned model.
4. Used a lightweight neural architecture search to find the best configuration for the pruned model.
5. Repeated this process iteratively to create even smaller models.
Cool, giving it a try this weekend 😎
Code: https://github.com/NVlabs/Minitron
Paper: https://arxiv.org/abs/2407.14679
Demo: nvidia/minitron
liked
2
models
6 months ago
Llama3.1 gguf model
4
#215 opened 6 months ago
by
ML-master-123
Important Information About this model (Q8)
1
#3 opened 6 months ago
by
karmikovic
liked
3
models
6 months ago
Very impressive!
2
#4 opened 7 months ago
by
DanielAdler666
liked
a
model
6 months ago
surprising - that model is incredible for it's size.
20
#2 opened 6 months ago
by
mirek190