Eurdem commited on
Commit
21d9b83
1 Parent(s): e9fe6c4

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +5 -35
README.md CHANGED
@@ -1,47 +1,17 @@
1
  ---
2
  license: apache-2.0
3
  tags:
4
- - moe
5
  - frankenmoe
6
  - merge
7
- - mergekit
8
- - lazymergekit
9
- - fblgit/UNA-TheBeagle-7b-v1
10
- - berkeley-nest/Starling-LM-7B-alpha
11
- base_model:
12
- - fblgit/UNA-TheBeagle-7b-v1
13
- - berkeley-nest/Starling-LM-7B-alpha
14
  ---
15
 
16
  # megatron_1.1_MoE_2x7B
17
 
18
- megatron_1.1_MoE_2x7B is a Mixure of Experts (MoE) made with the following models using [LazyMergekit](https://colab.research.google.com/drive/1obulZ1ROXHjYLn6PPZJwRR6GzgQogxxb?usp=sharing):
19
- * [fblgit/UNA-TheBeagle-7b-v1](https://huggingface.co/fblgit/UNA-TheBeagle-7b-v1)
20
- * [berkeley-nest/Starling-LM-7B-alpha](https://huggingface.co/berkeley-nest/Starling-LM-7B-alpha)
21
-
22
- ## 🧩 Configuration
23
-
24
- ```yaml
25
- base_model: openchat/openchat-3.5-0106
26
- gate_mode: hidden
27
- dtype: bfloat16
28
- experts:
29
- - source_model: fblgit/UNA-TheBeagle-7b-v1
30
- positive_prompts:
31
- - "Mathematics"
32
- - "Physics"
33
- negative_prompts:
34
- - "History"
35
- - "Philosophy"
36
- - source_model: berkeley-nest/Starling-LM-7B-alpha
37
- positive_prompts:
38
- - "Earth Sciences (Geology, Meteorology, Oceanography)"
39
- - "Environmental Science"
40
- negative_prompts:
41
- - "Education"
42
- - "Law"
43
 
44
- ```
45
 
46
  ## 💻 Usage
47
 
@@ -61,7 +31,7 @@ pipeline = transformers.pipeline(
61
  model_kwargs={"torch_dtype": torch.float16, "load_in_4bit": True},
62
  )
63
 
64
- messages = [{"role": "user", "content": "Explain what a Mixture of Experts is in less than 100 words."}]
65
  prompt = pipeline.tokenizer.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)
66
  outputs = pipeline(prompt, max_new_tokens=256, do_sample=True, temperature=0.7, top_k=50, top_p=0.95)
67
  print(outputs[0]["generated_text"])
 
1
  ---
2
  license: apache-2.0
3
  tags:
 
4
  - frankenmoe
5
  - merge
6
+ - MoE
7
+ - Mixtral
8
+
 
 
 
 
9
  ---
10
 
11
  # megatron_1.1_MoE_2x7B
12
 
13
+ megatron_1.1_MoE_2x7B is a Mixure of Experts (MoE) (mistral)
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
14
 
 
15
 
16
  ## 💻 Usage
17
 
 
31
  model_kwargs={"torch_dtype": torch.float16, "load_in_4bit": True},
32
  )
33
 
34
+ messages = [{"role": "user", "content": "Tell me about AI"}]
35
  prompt = pipeline.tokenizer.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)
36
  outputs = pipeline(prompt, max_new_tokens=256, do_sample=True, temperature=0.7, top_k=50, top_p=0.95)
37
  print(outputs[0]["generated_text"])