LoneStriker
commited on
Commit
•
d6e61d3
1
Parent(s):
f2ca433
Upload folder using huggingface_hub
Browse files- .gitattributes +9 -0
- Medorca-2x7b-Q3_K_L.gguf +3 -0
- Medorca-2x7b-Q3_K_M.gguf +3 -0
- Medorca-2x7b-Q3_K_S.gguf +3 -0
- Medorca-2x7b-Q4_K_M.gguf +3 -0
- Medorca-2x7b-Q4_K_S.gguf +3 -0
- Medorca-2x7b-Q5_K_M.gguf +3 -0
- Medorca-2x7b-Q5_K_S.gguf +3 -0
- Medorca-2x7b-Q6_K.gguf +3 -0
- Medorca-2x7b-Q8_0.gguf +3 -0
- README.md +94 -0
- huggingface-metadata.txt +10 -0
- mergekit_config.yml +17 -0
- mergekit_moe_config.yml +30 -0
.gitattributes
CHANGED
@@ -33,3 +33,12 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
|
|
33 |
*.zip filter=lfs diff=lfs merge=lfs -text
|
34 |
*.zst filter=lfs diff=lfs merge=lfs -text
|
35 |
*tfevents* filter=lfs diff=lfs merge=lfs -text
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
33 |
*.zip filter=lfs diff=lfs merge=lfs -text
|
34 |
*.zst filter=lfs diff=lfs merge=lfs -text
|
35 |
*tfevents* filter=lfs diff=lfs merge=lfs -text
|
36 |
+
Medorca-2x7b-Q3_K_L.gguf filter=lfs diff=lfs merge=lfs -text
|
37 |
+
Medorca-2x7b-Q3_K_M.gguf filter=lfs diff=lfs merge=lfs -text
|
38 |
+
Medorca-2x7b-Q3_K_S.gguf filter=lfs diff=lfs merge=lfs -text
|
39 |
+
Medorca-2x7b-Q4_K_M.gguf filter=lfs diff=lfs merge=lfs -text
|
40 |
+
Medorca-2x7b-Q4_K_S.gguf filter=lfs diff=lfs merge=lfs -text
|
41 |
+
Medorca-2x7b-Q5_K_M.gguf filter=lfs diff=lfs merge=lfs -text
|
42 |
+
Medorca-2x7b-Q5_K_S.gguf filter=lfs diff=lfs merge=lfs -text
|
43 |
+
Medorca-2x7b-Q6_K.gguf filter=lfs diff=lfs merge=lfs -text
|
44 |
+
Medorca-2x7b-Q8_0.gguf filter=lfs diff=lfs merge=lfs -text
|
Medorca-2x7b-Q3_K_L.gguf
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:5adc47244d850ff743298ed809993e90ff870f9cf4aa2e54dcb31e68d5049605
|
3 |
+
size 5829553280
|
Medorca-2x7b-Q3_K_M.gguf
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:0d751c74a73b27bf305dc8f66ff087e3ff57dadc7ed453b60f51cd5cd78045ca
|
3 |
+
size 5361364096
|
Medorca-2x7b-Q3_K_S.gguf
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:0c1a914d7e1f420fe2a1686e365b5403cf09904742eab4d76bdb45fbb5653e7c
|
3 |
+
size 4808764544
|
Medorca-2x7b-Q4_K_M.gguf
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:e71e2a3fcb1b7e78cdab94e19ed185f07a01c8a15e3504a89d781efc3697ea8f
|
3 |
+
size 6702338784
|
Medorca-2x7b-Q4_K_S.gguf
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:694c6ebd3ffc158703cc63047cdeeade03b6cef8a1e120a5ebff81b0592d66c8
|
3 |
+
size 6314627808
|
Medorca-2x7b-Q5_K_M.gguf
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:cb7d6d67c4de9b5795f437017ea65ecdfc86870a79ee3322307d7cebda40d039
|
3 |
+
size 7855380704
|
Medorca-2x7b-Q5_K_S.gguf
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:b2af7f642471e460dcd68b04a0baba0a3b693856b23eb60e957c67d8b8461aec
|
3 |
+
size 7628101856
|
Medorca-2x7b-Q6_K.gguf
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:893c13d7289e29840c670111f116d02ef228615fafcb81d2df5c4ed4589ec0ed
|
3 |
+
size 9080487744
|
Medorca-2x7b-Q8_0.gguf
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:5772b4c3970f3c07e4707ad5a8889fe40d970e0e011516b26793ad87fbdbd4b4
|
3 |
+
size 11760703104
|
README.md
ADDED
@@ -0,0 +1,94 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
---
|
2 |
+
license: apache-2.0
|
3 |
+
tags:
|
4 |
+
- moe
|
5 |
+
- merge
|
6 |
+
- epfl-llm/meditron-7b
|
7 |
+
- microsoft/Orca-2-7b
|
8 |
+
---
|
9 |
+
|
10 |
+
![image/png](https://cdn-uploads.huggingface.co/production/uploads/63486df1f8f01fcc4b23e97d/MVYcLAR1Inm5dY-XHiAhe.png)
|
11 |
+
|
12 |
+
# Medorca-2x7b
|
13 |
+
|
14 |
+
Medorca-2x7b is a Mixure of Experts (MoE) made with the following models:
|
15 |
+
* [epfl-llm/meditron-7b](https://huggingface.co/epfl-llm/meditron-7b)
|
16 |
+
* [microsoft/Orca-2-7b](https://huggingface.co/microsoft/Orca-2-7b)
|
17 |
+
|
18 |
+
## Evaluations
|
19 |
+
|
20 |
+
| Benchmark | Medorca-2x7b | Orca-2-7b | llama-2-7b | meditron-7b | meditron-70b |
|
21 |
+
| --- | --- | --- | --- | --- | --- |
|
22 |
+
| MedMCQA | | | | | |
|
23 |
+
| ClosedPubMedQA | | | | | |
|
24 |
+
| PubMedQA | | | | | |
|
25 |
+
| MedQA | | | | | |
|
26 |
+
| MedQA4 | | | | | |
|
27 |
+
| MedicationQA | | | | | |
|
28 |
+
| MMLU Medical | | | | | |
|
29 |
+
| MMLU | 53.3 | **56.37** | | | |
|
30 |
+
| TruthfulQA | 48.04 | **52.45** | | | |
|
31 |
+
| GSM8K | **20.64** | 14.71 | | | |
|
32 |
+
| ARC | 54.1 | 54.1 | | | |
|
33 |
+
| HellaSwag | 76.04 | **76.19** | | | |
|
34 |
+
| Winogrande | **74.51** | 73.48 | | | |
|
35 |
+
|
36 |
+
More details on the Open LLM Leaderboard evaluation results can be found [here](https://huggingface.co/datasets/open-llm-leaderboard/details_Technoculture__Medorca-2x7b.)
|
37 |
+
|
38 |
+
## 🧩 Configuration
|
39 |
+
|
40 |
+
```yaml
|
41 |
+
base_model: microsoft/Orca-2-7b
|
42 |
+
gate_mode: hidden
|
43 |
+
dtype: bfloat16
|
44 |
+
experts:
|
45 |
+
- source_model: epfl-llm/meditron-7b
|
46 |
+
positive_prompts:
|
47 |
+
- "How does sleep affect cardiovascular health?"
|
48 |
+
- "Could a plant-based diet improve arthritis symptoms?"
|
49 |
+
- "A patient comes in with symptoms of dizziness and nausea..."
|
50 |
+
- "When discussing diabetes management, the key factors to consider are..."
|
51 |
+
- "The differential diagnosis for a headache with visual aura could include..."
|
52 |
+
negative_prompts:
|
53 |
+
- "Recommend a good recipe for a vegetarian lasagna."
|
54 |
+
- "Give an overview of the French Revolution."
|
55 |
+
- "Explain how a digital camera captures an image."
|
56 |
+
- "What are the environmental impacts of deforestation?"
|
57 |
+
- "The recent advancements in artificial intelligence have led to developments in..."
|
58 |
+
- "The fundamental concepts in economics include ideas like supply and demand, which explain..."
|
59 |
+
- source_model: microsoft/Orca-2-7b
|
60 |
+
positive_prompts:
|
61 |
+
- "Here is a funny joke for you -"
|
62 |
+
- "When considering the ethical implications of artificial intelligence, one must take into account..."
|
63 |
+
- "In strategic planning, a company must analyze its strengths and weaknesses, which involves..."
|
64 |
+
- "Understanding consumer behavior in marketing requires considering factors like..."
|
65 |
+
- "The debate on climate change solutions hinges on arguments that..."
|
66 |
+
negative_prompts:
|
67 |
+
- "In discussing dietary adjustments for managing hypertension, it's crucial to emphasize..."
|
68 |
+
- "For early detection of melanoma, dermatologists recommend that patients regularly check their skin for..."
|
69 |
+
- "Explaining the importance of vaccination, a healthcare professional should highlight..."
|
70 |
+
```
|
71 |
+
|
72 |
+
## 💻 Usage
|
73 |
+
|
74 |
+
```python
|
75 |
+
!pip install -qU transformers bitsandbytes accelerate
|
76 |
+
|
77 |
+
from transformers import AutoTokenizer
|
78 |
+
import transformers
|
79 |
+
import torch
|
80 |
+
|
81 |
+
model = "Technoculture/Medorca-2x7b"
|
82 |
+
|
83 |
+
tokenizer = AutoTokenizer.from_pretrained(model)
|
84 |
+
pipeline = transformers.pipeline(
|
85 |
+
"text-generation",
|
86 |
+
model=model,
|
87 |
+
model_kwargs={"torch_dtype": torch.float16},
|
88 |
+
)
|
89 |
+
|
90 |
+
messages = [{"role": "user", "content": "Why am i feeling so tired this month?"}]
|
91 |
+
prompt = pipeline.tokenizer.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)
|
92 |
+
outputs = pipeline(prompt, max_new_tokens=256, do_sample=True, temperature=0.7, top_k=50, top_p=0.95)
|
93 |
+
print(outputs[0]["generated_text"])
|
94 |
+
```
|
huggingface-metadata.txt
ADDED
@@ -0,0 +1,10 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
url: https://huggingface.co/Technoculture/Medorca-2x7b
|
2 |
+
branch: main
|
3 |
+
download date: 2024-01-22 10:17:21
|
4 |
+
sha256sum:
|
5 |
+
b7312033024738baff24be5c914396422b027380cc3cbc44cb6c0eb62b74cf94 model-00001-of-00002.safetensors
|
6 |
+
f1c840e1ed0c60c73ab21953cf450dee5362a2412db3411fdae858d649d08261 model-00001-of-00003.safetensors
|
7 |
+
89ae0a8853bb7c1d3029f8a3bb295d8ba70835180159f5edd840fd28d24d947a model-00002-of-00002.safetensors
|
8 |
+
ae2432a8a64091cc18c79124794cdebdef9b2d0dfe6a6d466398f53e4d77e57e model-00002-of-00003.safetensors
|
9 |
+
24f8c33dd495ac7f6bb3210049edf02737817427591f78ce81257a78abad30bc model-00003-of-00003.safetensors
|
10 |
+
9e556afd44213b6bd1be2b850ebbbd98f5481437a8021afaf58ee7fb1818d347 tokenizer.model
|
mergekit_config.yml
ADDED
@@ -0,0 +1,17 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
|
2 |
+
slices:
|
3 |
+
- sources:
|
4 |
+
- model: epfl-llm/meditron-7b
|
5 |
+
layer_range: [0, 32]
|
6 |
+
- model: microsoft/Orca-2-7b
|
7 |
+
layer_range: [0, 32]
|
8 |
+
merge_method: slerp
|
9 |
+
base_model: epfl-llm/meditron-7b
|
10 |
+
parameters:
|
11 |
+
t:
|
12 |
+
- filter: self_attn
|
13 |
+
value: [0, 0.5, 0.3, 0.7, 1]
|
14 |
+
- filter: mlp
|
15 |
+
value: [1, 0.5, 0.7, 0.3, 0]
|
16 |
+
- value: 0.5 # fallback for rest of tensors
|
17 |
+
dtype: float16
|
mergekit_moe_config.yml
ADDED
@@ -0,0 +1,30 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
|
2 |
+
base_model: microsoft/Orca-2-7b
|
3 |
+
gate_mode: hidden
|
4 |
+
dtype: bfloat16
|
5 |
+
experts:
|
6 |
+
- source_model: epfl-llm/meditron-7b
|
7 |
+
positive_prompts:
|
8 |
+
- "How does sleep affect cardiovascular health?"
|
9 |
+
- "Could a plant-based diet improve arthritis symptoms?"
|
10 |
+
- "A patient comes in with symptoms of dizziness and nausea..."
|
11 |
+
- "When discussing diabetes management, the key factors to consider are..."
|
12 |
+
- "The differential diagnosis for a headache with visual aura could include..."
|
13 |
+
negative_prompts:
|
14 |
+
- "Recommend a good recipe for a vegetarian lasagna."
|
15 |
+
- "Give an overview of the French Revolution."
|
16 |
+
- "Explain how a digital camera captures an image."
|
17 |
+
- "What are the environmental impacts of deforestation?"
|
18 |
+
- "The recent advancements in artificial intelligence have led to developments in..."
|
19 |
+
- "The fundamental concepts in economics include ideas like supply and demand, which explain..."
|
20 |
+
- source_model: microsoft/Orca-2-7b
|
21 |
+
positive_prompts:
|
22 |
+
- "Here is a funny joke for you -"
|
23 |
+
- "When considering the ethical implications of artificial intelligence, one must take into account..."
|
24 |
+
- "In strategic planning, a company must analyze its strengths and weaknesses, which involves..."
|
25 |
+
- "Understanding consumer behavior in marketing requires considering factors like..."
|
26 |
+
- "The debate on climate change solutions hinges on arguments that..."
|
27 |
+
negative_prompts:
|
28 |
+
- "In discussing dietary adjustments for managing hypertension, it's crucial to emphasize..."
|
29 |
+
- "For early detection of melanoma, dermatologists recommend that patients regularly check their skin for..."
|
30 |
+
- "Explaining the importance of vaccination, a healthcare professional should highlight..."
|