Instructions to use QuixiAI/MegaDolphin-120b with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use QuixiAI/MegaDolphin-120b with Transformers:

# Use a pipeline as a high-level helper
from transformers import pipeline

pipe = pipeline("text-generation", model="QuixiAI/MegaDolphin-120b")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("QuixiAI/MegaDolphin-120b")
model = AutoModelForCausalLM.from_pretrained("QuixiAI/MegaDolphin-120b")
messages = [
    {"role": "user", "content": "Who are you?"},
]
inputs = tokenizer.apply_chat_template(
	messages,
	add_generation_prompt=True,
	tokenize=True,
	return_dict=True,
	return_tensors="pt",
).to(model.device)

outputs = model.generate(**inputs, max_new_tokens=40)
print(tokenizer.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use QuixiAI/MegaDolphin-120b with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "QuixiAI/MegaDolphin-120b"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "QuixiAI/MegaDolphin-120b",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/QuixiAI/MegaDolphin-120b

SGLang

How to use QuixiAI/MegaDolphin-120b with SGLang:

Install from pip and serve model

# Install SGLang from pip:
pip install sglang
# Start the SGLang server:
python3 -m sglang.launch_server \
    --model-path "QuixiAI/MegaDolphin-120b" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "QuixiAI/MegaDolphin-120b",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker images

docker run --gpus all \
    --shm-size 32g \
    -p 30000:30000 \
    -v ~/.cache/huggingface:/root/.cache/huggingface \
    --env "HF_TOKEN=<secret>" \
    --ipc=host \
    lmsysorg/sglang:latest \
    python3 -m sglang.launch_server \
        --model-path "QuixiAI/MegaDolphin-120b" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "QuixiAI/MegaDolphin-120b",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use QuixiAI/MegaDolphin-120b with Docker Model Runner:
```
docker model run hf.co/QuixiAI/MegaDolphin-120b
```

Update README.md

by Balajirajput966 - opened Aug 25, 2025

base: refs/heads/main

←

from: refs/pr/6

Discussion Files changed

+23

-8

Files changed (1) hide show

README.md +23 -8

README.md CHANGED Viewed

@@ -7,6 +7,9 @@ datasets:
 - jondurbin/airoboros-2.2.1
 - ehartford/samantha-data
 - ehartford/WizardLM_evol_instruct_V2_196k_unfiltered_merged_split
 base_model:
 - cognitivecomputations/dolphin-2.2-70b
 - cognitivecomputations/dolphin-2.2-70b
@@ -33,7 +36,8 @@ model-index:
       value: 69.03
       name: normalized accuracy
     source:
-      url: https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=cognitivecomputations/MegaDolphin-120b
       name: Open LLM Leaderboard
   - task:
       type: text-generation
@@ -49,7 +53,8 @@ model-index:
       value: 87.8
       name: normalized accuracy
     source:
-      url: https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=cognitivecomputations/MegaDolphin-120b
       name: Open LLM Leaderboard
   - task:
       type: text-generation
@@ -66,7 +71,8 @@ model-index:
       value: 69.26
       name: accuracy
     source:
-      url: https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=cognitivecomputations/MegaDolphin-120b
       name: Open LLM Leaderboard
   - task:
       type: text-generation
@@ -82,7 +88,8 @@ model-index:
     - type: mc2
       value: 59.28
     source:
-      url: https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=cognitivecomputations/MegaDolphin-120b
       name: Open LLM Leaderboard
   - task:
       type: text-generation
@@ -99,7 +106,8 @@ model-index:
       value: 81.85
       name: accuracy
     source:
-      url: https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=cognitivecomputations/MegaDolphin-120b
       name: Open LLM Leaderboard
   - task:
       type: text-generation
@@ -116,8 +124,16 @@ model-index:
       value: 46.25
       name: accuracy
     source:
-      url: https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=cognitivecomputations/MegaDolphin-120b
       name: Open LLM Leaderboard
 ---
 MegaDolphin 2.2 120b 🐬
@@ -416,5 +432,4 @@ Detailed results can be found [here](https://huggingface.co/datasets/open-llm-le
 |MMLU (5-Shot)                    |69.26|
 |TruthfulQA (0-shot)              |59.28|
 |Winogrande (5-shot)              |81.85|
-|GSM8k (5-shot)                   |46.25|

 - jondurbin/airoboros-2.2.1
 - ehartford/samantha-data
 - ehartford/WizardLM_evol_instruct_V2_196k_unfiltered_merged_split
+- fka/awesome-chatgpt-prompts
+- nvidia/Nemotron-Post-Training-Dataset-v1
+- frascuchon/fka_awesome-chatgpt-prompts__prompts.csv2
 base_model:
 - cognitivecomputations/dolphin-2.2-70b
 - cognitivecomputations/dolphin-2.2-70b
       value: 69.03
       name: normalized accuracy
     source:
+      url: >-
+        https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=cognitivecomputations/MegaDolphin-120b
       name: Open LLM Leaderboard
   - task:
       type: text-generation
       value: 87.8
       name: normalized accuracy
     source:
+      url: >-
+        https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=cognitivecomputations/MegaDolphin-120b
       name: Open LLM Leaderboard
   - task:
       type: text-generation
       value: 69.26
       name: accuracy
     source:
+      url: >-
+        https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=cognitivecomputations/MegaDolphin-120b
       name: Open LLM Leaderboard
   - task:
       type: text-generation
     - type: mc2
       value: 59.28
     source:
+      url: >-
+        https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=cognitivecomputations/MegaDolphin-120b
       name: Open LLM Leaderboard
   - task:
       type: text-generation
       value: 81.85
       name: accuracy
     source:
+      url: >-
+        https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=cognitivecomputations/MegaDolphin-120b
       name: Open LLM Leaderboard
   - task:
       type: text-generation
       value: 46.25
       name: accuracy
     source:
+      url: >-
+        https://huggingface.co/spaces/HuggingFaceH4/open_llm_leaderboard?query=cognitivecomputations/MegaDolphin-120b
       name: Open LLM Leaderboard
+metrics:
+- accuracy
+new_version: google/gemma-3-270m
+pipeline_tag: any-to-any
+library_name: adapter-transformers
+tags:
+- biology
 ---
 MegaDolphin 2.2 120b 🐬
 |MMLU (5-Shot)                    |69.26|
 |TruthfulQA (0-shot)              |59.28|
 |Winogrande (5-shot)              |81.85|
+|GSM8k (5-shot)                   |46.25|