Views
No views yet
| Name | Quant method | Size |
|---|---|---|
| mixtral-gqa-400m-v2.fp16.gguf | fp16 | 4.01 GB |
| mixtral-gqa-400m-v2.q2_k.gguf | q2_k | 703.28 MB |
| mixtral-gqa-400m-v2.q3_k_m.gguf | q3_k_m | 899.86 MB |
| mixtral-gqa-400m-v2.q4_k_m.gguf | q4_k_m | 1.15 GB |
| mixtral-gqa-400m-v2.q5_k_m.gguf | q5_k_m | 1.39 GB |
| mixtral-gqa-400m-v2.q6_k.gguf | q6_k | 1.65 GB |
| mixtral-gqa-400m-v2.q8_0.gguf | q8_0 | 2.13 GB |
1# !pip install -U -q transformers datasets accelerate sentencepiece
2import pprint as pp
3from transformers import pipeline
4
5pipe = pipeline(
6 "text-generation",
7 model="BEE-spoke-data/Mixtral-GQA-400m-v2",
8 device_map="auto",
9)
10pipe.model.config.pad_token_id = pipe.model.config.eos_token_id
11
12prompt = "My favorite movie is Godfather because"
13
14res = pipe(
15 prompt,
16 max_new_tokens=256,
17 top_k=4,
18 penalty_alpha=0.6,
19 use_cache=True,
20 no_repeat_ngram_size=4,
21 repetition_penalty=1.1,
22 renormalize_logits=True,
23)
24pp.pprint(res[0])