Instructions to use arcee-ai/AFM-4.5B-Preview with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use arcee-ai/AFM-4.5B-Preview with Transformers:

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

pipe = pipeline("text-generation", model="arcee-ai/AFM-4.5B-Preview")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("arcee-ai/AFM-4.5B-Preview")
model = AutoModelForCausalLM.from_pretrained("arcee-ai/AFM-4.5B-Preview")
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

vLLM

How to use arcee-ai/AFM-4.5B-Preview with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "arcee-ai/AFM-4.5B-Preview"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "arcee-ai/AFM-4.5B-Preview",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/arcee-ai/AFM-4.5B-Preview

SGLang

How to use arcee-ai/AFM-4.5B-Preview 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 "arcee-ai/AFM-4.5B-Preview" \
    --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": "arcee-ai/AFM-4.5B-Preview",
		"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 "arcee-ai/AFM-4.5B-Preview" \
        --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": "arcee-ai/AFM-4.5B-Preview",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use arcee-ai/AFM-4.5B-Preview with Docker Model Runner:
```
docker model run hf.co/arcee-ai/AFM-4.5B-Preview
```

MaziyarPanahi commited on Aug 2, 2025

Commit

71d9f4e

verified ·

1 Parent(s): 8035825

Upload folder using huggingface_hub

Browse files

Files changed (4) hide show

config.json +1 -6
generation_config.json +1 -1
model.safetensors.index.json +1 -0
tokenizer_config.json +1 -2

config.json CHANGED Viewed

@@ -4,11 +4,6 @@
   ],
   "attention_bias": false,
   "attention_dropout": 0.0,
-  "auto_map": {
-    "AutoConfig": "MaziyarPanahi/afm-sage-v8.4--configuration_arcee.ArceeConfig",
-    "AutoModel": "MaziyarPanahi/afm-sage-v8.4--modeling_arcee.ArceeForCausalLM",
-    "AutoModelForCausalLM": "MaziyarPanahi/afm-sage-v8.4--modeling_arcee.ArceeForCausalLM"
-  },
   "bos_token_id": 128000,
   "eos_token_id": 128003,
   "head_dim": 128,
@@ -35,7 +30,7 @@
   "rope_theta": 10000.0,
   "tie_word_embeddings": false,
   "torch_dtype": "bfloat16",
-  "transformers_version": "4.52.4",
   "use_cache": false,
   "vocab_size": 128064
 }

   ],
   "attention_bias": false,
   "attention_dropout": 0.0,
   "bos_token_id": 128000,
   "eos_token_id": 128003,
   "head_dim": 128,
   "rope_theta": 10000.0,
   "tie_word_embeddings": false,
   "torch_dtype": "bfloat16",
+  "transformers_version": "4.54.1",
   "use_cache": false,
   "vocab_size": 128064
 }

generation_config.json CHANGED Viewed

@@ -2,6 +2,6 @@
   "_from_model_config": true,
   "bos_token_id": 128000,
   "eos_token_id": 128003,
-  "transformers_version": "4.52.4",
   "use_cache": false
 }

   "_from_model_config": true,
   "bos_token_id": 128000,
   "eos_token_id": 128003,
+  "transformers_version": "4.54.1",
   "use_cache": false
 }

model.safetensors.index.json CHANGED Viewed

@@ -1,5 +1,6 @@
 {
   "metadata": {
     "total_size": 9238983680
   },
   "weight_map": {

 {
   "metadata": {
+    "total_parameters": 4619491840,
     "total_size": 9238983680
   },
   "weight_map": {

tokenizer_config.json CHANGED Viewed

@@ -42,7 +42,6 @@
     }
   },
   "bos_token": "<|begin_of_text|>",
-  "chat_template": "{{ bos_token }}{%- set loop_messages = messages %}\n{%- if loop_messages[0]['role'] != 'system' %}\n<|im_start|>system\nYou are a helpful assistant named AFM-4.5B-Preview, trained by Arcee AI.<|im_end|>\n{%- endif %}\n{%- for message in loop_messages %}\n{%- set content = '<|im_start|>' + message['role'] + '\\n'+ message['content'] | trim %}\n{%- if loop.index0 == 0 %}\n{%- set content = content %}\n{%- endif %}\n{%- if not (loop.last and message['role'] == 'assistant') %}\n{%- set content = content + '<|im_end|>\\n' %}\n{%- endif %}\n{{- content }}\n{%- endfor %}\n{%- if messages[-1]['role'] != 'assistant' %}\n{{- '<|im_start|>assistant\\n' }}\n{%- endif %}",
   "clean_up_tokenization_spaces": true,
   "eos_token": "<|end_of_text|>",
   "extra_special_tokens": {},
@@ -53,5 +52,5 @@
   "model_max_length": 65536,
   "pad_token": "<|finetune_right_pad_id|>",
   "padding_side": "left",
-  "tokenizer_class": "PreTrainedTokenizer"
 }

     }
   },
   "bos_token": "<|begin_of_text|>",
   "clean_up_tokenization_spaces": true,
   "eos_token": "<|end_of_text|>",
   "extra_special_tokens": {},
   "model_max_length": 65536,
   "pad_token": "<|finetune_right_pad_id|>",
   "padding_side": "left",
+  "tokenizer_class": "PreTrainedTokenizerFast"
 }