Views
No views yet
| Name | Quant method | Size |
|---|---|---|
| MiniMerlin-3B.Q2_K.gguf | Q2_K | 1.09GB |
| MiniMerlin-3B.IQ3_XS.gguf | IQ3_XS | 1.21GB |
| MiniMerlin-3B.IQ3_S.gguf | IQ3_S | 1.27GB |
| MiniMerlin-3B.Q3_K_S.gguf | Q3_K_S | 1.27GB |
| MiniMerlin-3B.IQ3_M.gguf | IQ3_M | 1.33GB |
| MiniMerlin-3B.Q3_K.gguf | Q3_K | 1.4GB |
| MiniMerlin-3B.Q3_K_M.gguf | Q3_K_M | 1.4GB |
| MiniMerlin-3B.Q3_K_L.gguf | Q3_K_L | 1.52GB |
| MiniMerlin-3B.IQ4_XS.gguf | IQ4_XS | 1.55GB |
| MiniMerlin-3B.Q4_0.gguf | Q4_0 | 1.62GB |
| MiniMerlin-3B.IQ4_NL.gguf | IQ4_NL | 1.63GB |
| MiniMerlin-3B.Q4_K_S.gguf | Q4_K_S | 1.63GB |
| MiniMerlin-3B.Q4_K.gguf | Q4_K | 1.72GB |
| MiniMerlin-3B.Q4_K_M.gguf | Q4_K_M | 1.72GB |
| MiniMerlin-3B.Q4_1.gguf | Q4_1 | 1.79GB |
| MiniMerlin-3B.Q5_0.gguf | Q5_0 | 1.95GB |
| MiniMerlin-3B.Q5_K_S.gguf | Q5_K_S | 1.95GB |
| MiniMerlin-3B.Q5_K.gguf | Q5_K | 2.01GB |
| MiniMerlin-3B.Q5_K_M.gguf | Q5_K_M | 2.01GB |
| MiniMerlin-3B.Q5_1.gguf | Q5_1 | 2.12GB |
| MiniMerlin-3B.Q6_K.gguf | Q6_K | 2.31GB |
| MiniMerlin-3B.Q8_0.gguf | Q8_0 | 2.99GB |
1from transformers import AutoModelForCausalLM, AutoTokenizer
2from peft import PeftModel
3import torch
4
5model = AutoModelForCausalLM.from_pretrained(
6 "teilomillet/MiniMerlin-3B",
7 revision="0.1",
8 return_dict=True,
9 torch_dtype=torch.bfloat16,
10 device_map='auto'
11)
12
13tokenizer = AutoTokenizer.from_pretrained("teilomillet/MiniMerlin-3B")
14tokenizer.pad_token = tokenizer.eos_token
15
16text = "[|User|] Comment faire un bon plat ? </s>[|Assistant|]"
17inputs = tokenizer(text, return_tensors="pt").to(0)
18
19outputs = model.generate(**inputs, max_new_tokens=800)
20print(tokenizer.decode(outputs[0], skip_special_tokens=False))