Instructions to use IFM/K2-Horizon-375B-A23B with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use IFM/K2-Horizon-375B-A23B with Transformers:

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

pipe = pipeline("text-generation", model="IFM/K2-Horizon-375B-A23B", trust_remote_code=True)
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoModelForCausalLM
model = AutoModelForCausalLM.from_pretrained("IFM/K2-Horizon-375B-A23B", trust_remote_code=True, device_map="auto")

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use IFM/K2-Horizon-375B-A23B with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "IFM/K2-Horizon-375B-A23B"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "IFM/K2-Horizon-375B-A23B",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/IFM/K2-Horizon-375B-A23B

SGLang

How to use IFM/K2-Horizon-375B-A23B 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 "IFM/K2-Horizon-375B-A23B" \
    --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": "IFM/K2-Horizon-375B-A23B",
		"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 "IFM/K2-Horizon-375B-A23B" \
        --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": "IFM/K2-Horizon-375B-A23B",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use IFM/K2-Horizon-375B-A23B with Docker Model Runner:
```
docker model run hf.co/IFM/K2-Horizon-375B-A23B
```

desaifan-mbzuai commited on 4 days ago

Commit

828b8de

verified ·

1 Parent(s): 6e4d9ce

Add Training Overview, Release Artifacts and Checkpoint Inventory sections to model card

Browse files

Files changed (1) hide show

README.md +76 -0

README.md CHANGED Viewed

@@ -120,6 +120,82 @@ outputs = model.generate(**inputs, max_new_tokens=32768, temperature=1.0, top_p=
 print(tokenizer.decode(outputs[0], skip_special_tokens=True))
 ```
 ## Best Practices
 1. **Reasoning effort: always `high`.** All reported results use high reasoning effort. Pass `{"chat_template_kwargs": {"reasoning_effort": "high"}}` on every request.

 print(tokenizer.decode(outputs[0], skip_special_tokens=True))
 ```
+## Training Overview
+The table describes the training stages in order, and the general purpose for the overall training approach.
+| Training stage | Training steps | Token budget (tokens) | Purpose |
+| --- | --- | --- | --- |
+| Pretraining — Phase 1 | 211000 | 7T   | Pretraining. |
+| Pretraining — Phase 2 | 240000 | 8T   | Pretraining, with newer and better data. |
+| Midtraining — Stage 1 | 32500  | 1T   | Context extension. |
+| Midtraining — Stage 2 | 15000  | 500B | Context extension. |
+| Midtraining — Stage 3 | 3500   | 120B | Context extension. |
+| Midtraining — Stage 4 | 6000   | 200B | Context extension. |
+| RL | N/A | N/A | We trained five experts starting from end of Midtraining — Stage 4, including knowledge work expert, IF expert, search expert, tool use expert, and reasoning expert. We then merged them. |
+| SFT — Phase 1 | 2400 | 80B  | Alignment. |
+| SFT — Phase 2 | 6000 | 200B | Alignment. |
+| SFT — Phase 3 | 1500 | 50B  | Alignment. |
+## Release Artifacts
+The tables below list the release artifacts for **K2-Horizon-375B-A23B**, their availability, and the expected release dates for remaining items.
+**Last updated:** [09-10-2026]
+**Status:**
+ - **Available** — fully released for the scope listed;
+ - **Partial** — some items are available, with remaining items listed in the notes;
+ - **In Progress** — intended for release but not yet available.
+### Artifact Index
+| Artifact | Link | Status | Remaining items / expected availability |
+| --- | --- | --- | --- |
+| Model card | [Hugging Face](https://huggingface.co/IFM/K2-Horizon-375B-A23B) | Available | N/A |
+| Training logs | [W&B](https://wandb.ai/llm360/K2-Horizon-375B) | Available | N/A |
+| Blog post | [Blog URL](https://ifm.ai/blog/k2/) | Available | N/A |
+| Checkpoints | [Checkpoint inventory](#checkpoint-inventory) | Partial | See details below |
+| Technical report | N/A | In Progress | End of September |
+| Code repository | [Github](https://github.com/ifm-ai/xllm) | In Progress | End of September |
+### Checkpoint Inventory
+**Model repository:** [IFM/K2-Horizon-375B-A23B](https://huggingface.co/IFM/K2-Horizon-375B-A23B)
+Branch names below refer to this repository. Patterns containing `*` describe branch naming conventions; intermediate checkpoints are listed in groups while the final checkpoint is listed separately. For a partially released group, the available checkpoints and the remaining checkpoints are listed in the notes.
+| Checkpoint | Branch / repository | Status | Remaining items / expected availability |
+| --- | --- | --- | --- |
+| Pretrain Phase 1 Intermediate Checkpoints | `pretrain_ph1_*` | Available | N/A |
+| Pretrain Phase 1 Final Checkpoint | `pretrain_ph1_211000` | Available | N/A |
+| Pretrain Phase 2 Intermediate Checkpoints | `pretrain_ph2_*` | Available | N/A |
+| Pretrain Phase 2 Final Checkpoint | `pretrain_ph2_240000` | Available | N/A |
+| Midtrain Stage 1 Intermediate Checkpoints | `mid_1_*` | Available | N/A |
+| Midtrain Stage 1 Final Checkpoint | `mid_1_32500` | Available | N/A |
+| Midtrain Stage 2 Intermediate Checkpoints | `mid_2_*` | Available | N/A |
+| Midtrain Stage 2 Final Checkpoint | `mid_2_15000` | Available | N/A |
+| Midtrain Stage 3 Intermediate Checkpoints | `mid_3_*` | Available | N/A |
+| Midtrain Stage 3 Final Checkpoint | `mid_3_3500` | Available | N/A |
+| Midtrain Stage 4 Intermediate Checkpoints | `mid_4_*` | Available | N/A |
+| Midtrain Stage 4 Final Checkpoint | `mid_4_6000` | Available | N/A |
+| RL Knowledge Work Expert Checkpoint | `rl_knowledge_work` | In Progress | Mid September |
+| RL IF Expert Checkpoint | `rl_if` | In Progress | Mid September |
+| RL Search Expert Checkpoint | `rl_search` | In Progress | Mid September |
+| RL Tool Use Expert Checkpoint | `rl_tool_use` | In Progress | Mid September |
+| RL Reasoning Expert Checkpoint | `rl_reasoning` | In Progress | Mid September |
+| RL Merged Final Checkpoint | `rl_merged` | Available | N/A |
+| SFT Phase 1 Intermediate Checkpoints | `sft_1_*` | Available | N/A |
+| SFT Phase 1 Final Checkpoint | `sft_1_2400` | Available | N/A |
+| SFT Phase 2 Intermediate Checkpoints | `sft_2_*` | Available | N/A |
+| SFT Phase 2 Final Checkpoint | `sft_2_6000` | Available | N/A |
+| SFT Phase 3 Intermediate Checkpoints | `sft_3_*` | Available | N/A |
+| SFT Phase 3 Final Checkpoint | `sft_3_1500` | Available | N/A |
 ## Best Practices
 1. **Reasoning effort: always `high`.** All reported results use high reasoning effort. Pass `{"chat_template_kwargs": {"reasoning_effort": "high"}}` on every request.