Instructions to use IFM/K2-Horizon-0.9B with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use IFM/K2-Horizon-0.9B with Transformers:

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

pipe = pipeline("text-generation", model="IFM/K2-Horizon-0.9B", trust_remote_code=True)
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# pip install -U transformers accelerate
# Load model directly
from transformers import AutoModelForCausalLM
model = AutoModelForCausalLM.from_pretrained("IFM/K2-Horizon-0.9B", trust_remote_code=True, device_map="auto")

Inference
Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use IFM/K2-Horizon-0.9B with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "IFM/K2-Horizon-0.9B"
# 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-0.9B",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/IFM/K2-Horizon-0.9B

SGLang

How to use IFM/K2-Horizon-0.9B 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-0.9B" \
    --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-0.9B",
		"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-0.9B" \
        --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-0.9B",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use IFM/K2-Horizon-0.9B with Docker Model Runner:
```
docker model run hf.co/IFM/K2-Horizon-0.9B
```

K2-Horizon-0.9B

Commit History

Nov 4 Yesterday

03:08
Update README.md

hunterhector

Verified signature
c4f92a0

Nov 2 3 days ago

19:04
Fix `output_hidden_states` / `attentions` returning None (missing `@capture_outputs`) (#8)⁠

zhou94539 +1: Avicennasis

9f3d124

Oct 26 10 days ago

02:52
Fix: add the missing @capture_outputs decorator (output_hidden_states is None)⁠

chen11003 +1: Avicennasis

d50881f

Oct 20 16 days ago

02:54
Align chat_template.jinja (#15)⁠

zhou94539 +1: aaryamonvikram

7c75f2f
02:53
Removal of redundant chat templates (#14)⁠

zhou94539 +1: aaryamonvikram

f7bbede

Oct 6 30 days ago

06:41
Support OpenAI-style text content lists (#10)⁠

hunterhector +1: hanseungwook

383cb66

Oct 1

16:45
Promote checkpoint rl-mopd_249 to main

HF release automation

76db896
15:46
Fix typo and wording.

desaifan-mbzuai

Verified signature
108340d

Sep 28

16:11
Release artifacts updated.

desaifan-mbzuai

Verified signature
fa7f5de

Sep 24

20:11
Delete files migration_manifest.json validation.json with huggingface_hub

zhou94539

Verified signature
6fbc54f

Sep 22

23:27
Add a note on ckpt usage

desaifan-mbzuai

Verified signature
309fc17

Sep 21

16:48
Update model card

zhou94539

Verified signature
6436b72
01:40
Update model card

zhou94539

Verified signature
a9e364e

Sep 18

00:05
Preserve checkpoint history: rl_stem⁠

HF release automation

9fa6faa
00:05
Publish checkpoint rl_stem⁠

HF release automation

acd705e
00:05
Preserve checkpoint history: rl_if⁠

HF release automation

ddbe3c9
00:05
Publish checkpoint rl_if⁠

HF release automation

be908c0
00:04
Preserve checkpoint history: rl_code2⁠

HF release automation

fb59197
00:04
Publish checkpoint rl_code2⁠

HF release automation

49b919a
00:04
Preserve checkpoint history: rl_math2b⁠

HF release automation

69bcdbf
00:04
Publish checkpoint rl_math2b⁠

HF release automation

c1e52cd
00:01
Preserve checkpoint history: rl_math2a⁠

HF release automation

6fa6461
00:01
Publish checkpoint rl_math2a⁠

HF release automation

0d0a126

Sep 17

23:58
Preserve checkpoint history: rl_code1⁠

HF release automation

092a2c5
23:58
Publish checkpoint rl_code1⁠

HF release automation

8c9b1ad
23:58
Preserve checkpoint history: rl_math1⁠

HF release automation

ce6e8d0
23:58
Publish checkpoint rl_math1⁠

HF release automation

f7c200e

Sep 16

22:42
Update ckpt progress

desaifan-mbzuai

Verified signature
d0a0fa2

Sep 15

22:27
Update README.md⁠

zhou94539

Verified signature
40b7742
21:03
Fix words to be accurate

desaifan-mbzuai

Verified signature
c58113b
17:55
Preserve checkpoint history: rl_rl-merged⁠

HF release automation

6ac3c9f
17:55
Preserve checkpoint history: rl-mopd_249⁠

HF release automation

312356f
17:55
Preserve checkpoint history: mid_2_5000⁠

HF release automation

08a5947
17:55
Preserve checkpoint history: mid_2_10000⁠

HF release automation

95607e8
17:55
Preserve checkpoint history: mid_2_15000⁠

HF release automation

763a1d2
17:55
Preserve checkpoint history: mid_2_20000⁠

HF release automation

6da49fa
17:55
Preserve checkpoint history: mid_2_25000⁠

HF release automation

7988f4c
17:55
Preserve checkpoint history: mid_2_30000⁠

HF release automation

3048fc5
17:55
Preserve checkpoint history: mid_2_35000⁠

HF release automation

43a22e8
17:55
Preserve checkpoint history: mid_2_40000⁠

HF release automation

effff8f
17:55
Preserve checkpoint history: mid_2_45000⁠

HF release automation

125fb36
17:55
Preserve checkpoint history: mid_2_47684⁠

HF release automation

75e9db5
17:55
Preserve checkpoint history: mid_1_50000⁠

HF release automation

aed6997
17:55
Preserve checkpoint history: mid_1_55000⁠

HF release automation

9b5121a
17:55
Preserve checkpoint history: mid_1_60000⁠

HF release automation

6a8c7f1
17:55
Preserve checkpoint history: mid_1_65000⁠

HF release automation

2238bf2
17:55
Preserve checkpoint history: mid_1_70000⁠

HF release automation

d6c4954
17:55
Preserve checkpoint history: mid_1_75000⁠

HF release automation

3feb7e1
17:55
Preserve checkpoint history: pretrain_500000⁠

HF release automation

757516c
17:55
Preserve checkpoint history: pretrain_510000⁠

HF release automation

4356ff4

Commit History

Nov 4 Yesterday

Update README.md

Nov 2 3 days ago

Fix `output_hidden_states` / `attentions` returning None (missing `@capture_outputs`) (#8)⁠

Oct 26 10 days ago

Fix: add the missing @capture_outputs decorator (output_hidden_states is None)⁠

Oct 20 16 days ago

Align chat_template.jinja (#15)⁠

Removal of redundant chat templates (#14)⁠

Oct 6 30 days ago

Support OpenAI-style text content lists (#10)⁠

Oct 1

Promote checkpoint rl-mopd_249 to main

Fix typo and wording.

Sep 28

Release artifacts updated.

Sep 24

Delete files migration_manifest.json validation.json with huggingface_hub

Sep 22

Add a note on ckpt usage

Sep 21

Update model card

Update model card

Sep 18

Preserve checkpoint history: rl_stem⁠

Publish checkpoint rl_stem⁠

Preserve checkpoint history: rl_if⁠

Publish checkpoint rl_if⁠

Preserve checkpoint history: rl_code2⁠

Publish checkpoint rl_code2⁠

Preserve checkpoint history: rl_math2b⁠

Publish checkpoint rl_math2b⁠

Preserve checkpoint history: rl_math2a⁠

Publish checkpoint rl_math2a⁠

Sep 17

Preserve checkpoint history: rl_code1⁠

Publish checkpoint rl_code1⁠

Preserve checkpoint history: rl_math1⁠

Publish checkpoint rl_math1⁠

Sep 16

Update ckpt progress

Sep 15

Update README.md⁠

Fix words to be accurate

Preserve checkpoint history: rl_rl-merged⁠

Preserve checkpoint history: rl-mopd_249⁠

Preserve checkpoint history: mid_2_5000⁠

Preserve checkpoint history: mid_2_10000⁠

Preserve checkpoint history: mid_2_15000⁠

Preserve checkpoint history: mid_2_20000⁠

Preserve checkpoint history: mid_2_25000⁠

Preserve checkpoint history: mid_2_30000⁠

Preserve checkpoint history: mid_2_35000⁠

Preserve checkpoint history: mid_2_40000⁠

Preserve checkpoint history: mid_2_45000⁠

Preserve checkpoint history: mid_2_47684⁠

Preserve checkpoint history: mid_1_50000⁠

Preserve checkpoint history: mid_1_55000⁠

Preserve checkpoint history: mid_1_60000⁠

Preserve checkpoint history: mid_1_65000⁠

Preserve checkpoint history: mid_1_70000⁠

Preserve checkpoint history: mid_1_75000⁠

Preserve checkpoint history: pretrain_500000⁠

Preserve checkpoint history: pretrain_510000⁠