Instructions to use trl-internal-testing/tiny-Lfm2ForCausalLM-2.5 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use trl-internal-testing/tiny-Lfm2ForCausalLM-2.5 with Transformers:

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

pipe = pipeline("text-generation", model="trl-internal-testing/tiny-Lfm2ForCausalLM-2.5")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# pip install -U transformers accelerate
# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("trl-internal-testing/tiny-Lfm2ForCausalLM-2.5")
model = AutoModelForCausalLM.from_pretrained("trl-internal-testing/tiny-Lfm2ForCausalLM-2.5", device_map="auto")
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=256)
print(tokenizer.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use trl-internal-testing/tiny-Lfm2ForCausalLM-2.5 with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "trl-internal-testing/tiny-Lfm2ForCausalLM-2.5"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "trl-internal-testing/tiny-Lfm2ForCausalLM-2.5",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/trl-internal-testing/tiny-Lfm2ForCausalLM-2.5

SGLang

How to use trl-internal-testing/tiny-Lfm2ForCausalLM-2.5 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 "trl-internal-testing/tiny-Lfm2ForCausalLM-2.5" \
    --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": "trl-internal-testing/tiny-Lfm2ForCausalLM-2.5",
		"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 "trl-internal-testing/tiny-Lfm2ForCausalLM-2.5" \
        --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": "trl-internal-testing/tiny-Lfm2ForCausalLM-2.5",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use trl-internal-testing/tiny-Lfm2ForCausalLM-2.5 with Docker Model Runner:
```
docker model run hf.co/trl-internal-testing/tiny-Lfm2ForCausalLM-2.5
```

Upload Lfm2ForCausalLM

by albertvillanova HF Staff - opened 1 day ago

base: refs/heads/main

←

from: refs/pr/1

Discussion Files changed

+25

-4

Files changed (3) hide show

chat_template.jinja +4 -2
config.json +20 -1
model.safetensors +1 -1

chat_template.jinja CHANGED Viewed

@@ -75,8 +75,10 @@
     {{- "<|im_start|>" + message.role + "\n" -}}
     {%- if message.role == "assistant" -%}
         {%- generation -%}
-        {%- if message.thinking is defined and (preserve_thinking or loop.index0 > ns.last_user_index) -%}
-            {{- "<think>" + message.thinking + "</think>" -}}
         {%- endif -%}
         {%- set _cfm_tag = "CONTINUE_FINAL_MESSAGE_TAG " -%}
         {%- set _has_cfm = false -%}

     {{- "<|im_start|>" + message.role + "\n" -}}
     {%- if message.role == "assistant" -%}
         {%- generation -%}
+        {%- set thinking = message.thinking or message.reasoning or message.reasoning_content -%}
+        {%- set thinking = thinking if thinking is string else "" -%}
+        {%- if thinking and (preserve_thinking or loop.index0 > ns.last_user_index) -%}
+            {{- "<think>" + thinking + "</think>" -}}
         {%- endif -%}
         {%- set _cfm_tag = "CONTINUE_FINAL_MESSAGE_TAG " -%}
         {%- set _has_cfm = false -%}

config.json CHANGED Viewed

@@ -2,14 +2,29 @@
   "architectures": [
     "Lfm2ForCausalLM"
   ],
   "block_auto_adjust_ff_dim": false,
   "block_ffn_dim_multiplier": 1.0,
   "block_multiple_of": 256,
   "bos_token_id": 1,
   "conv_L_cache": 3,
   "conv_bias": false,
   "dtype": "bfloat16",
   "eos_token_id": 7,
   "hidden_size": 8,
   "initializer_range": 0.02,
   "intermediate_size": 32,
@@ -21,6 +36,7 @@
   "model_type": "lfm2",
   "norm_eps": 1e-05,
   "num_attention_heads": 4,
   "num_hidden_layers": 2,
   "num_key_value_heads": 2,
   "pad_token_id": 0,
@@ -28,8 +44,11 @@
     "rope_theta": 1000000.0,
     "rope_type": "default"
   },
   "tie_word_embeddings": true,
   "transformers_version": "5.0.0",
-  "use_cache": false,
   "vocab_size": 65536
 }

   "architectures": [
     "Lfm2ForCausalLM"
   ],
+  "block__name_mlp": "parallel_mlp_merged",
   "block_auto_adjust_ff_dim": false,
+  "block_dim": 8,
+  "block_ff_dim": 32,
   "block_ffn_dim_multiplier": 1.0,
+  "block_ffn_te_autocast": false,
+  "block_ffn_use_quantized_params": false,
+  "block_mlp_init_scale": 1.0,
   "block_multiple_of": 256,
+  "block_norm_eps": 1e-05,
+  "block_out_init_scale": 1.0,
+  "block_sequence_parallel_norm_across_tp": false,
+  "block_use_swiglu": true,
+  "block_use_xavier_init": true,
   "bos_token_id": 1,
   "conv_L_cache": 3,
   "conv_bias": false,
+  "conv_dim": 8,
+  "conv_use_xavier_init": true,
   "dtype": "bfloat16",
   "eos_token_id": 7,
+  "ffn_te_autocast": false,
+  "ffn_use_quantized_params": false,
   "hidden_size": 8,
   "initializer_range": 0.02,
   "intermediate_size": 32,
   "model_type": "lfm2",
   "norm_eps": 1e-05,
   "num_attention_heads": 4,
+  "num_heads": 4,
   "num_hidden_layers": 2,
   "num_key_value_heads": 2,
   "pad_token_id": 0,
     "rope_theta": 1000000.0,
     "rope_type": "default"
   },
+  "sequence_parallel_norm_across_tp": false,
+  "tie_embedding": true,
   "tie_word_embeddings": true,
   "transformers_version": "5.0.0",
+  "use_cache": true,
+  "use_pos_enc": true,
   "vocab_size": 65536
 }

model.safetensors CHANGED Viewed

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:ce2d332095a2ca14ad26573a9b50a8be2a0445be817c187e5b9701b347aadbaa
 size 1054864

 version https://git-lfs.github.com/spec/v1
+oid sha256:75b4a6eb993c7f2777ee0621be1e667a72bfeee9b23bc3ede333c45135ab668d
 size 1054864