Instructions to use habdine/Esm2Text-Base-v1-1 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use habdine/Esm2Text-Base-v1-1 with Transformers:

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

pipe = pipeline("text-generation", model="habdine/Esm2Text-Base-v1-1", trust_remote_code=True)

# pip install -U transformers accelerate
# Load model directly
from transformers import AutoModel
model = AutoModel.from_pretrained("habdine/Esm2Text-Base-v1-1", trust_remote_code=True, device_map="auto")

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use habdine/Esm2Text-Base-v1-1 with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "habdine/Esm2Text-Base-v1-1"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "habdine/Esm2Text-Base-v1-1",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker

docker model run hf.co/habdine/Esm2Text-Base-v1-1

SGLang

How to use habdine/Esm2Text-Base-v1-1 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 "habdine/Esm2Text-Base-v1-1" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "habdine/Esm2Text-Base-v1-1",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

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 "habdine/Esm2Text-Base-v1-1" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "habdine/Esm2Text-Base-v1-1",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Docker Model Runner
How to use habdine/Esm2Text-Base-v1-1 with Docker Model Runner:
```
docker model run hf.co/habdine/Esm2Text-Base-v1-1
```

habdine commited on Oct 21, 2024

Commit

71ad8b3

verified ·

1 Parent(s): b5bd029

Update modeling_prot2text.py

Browse files

Files changed (1) hide show

modeling_prot2text.py +28 -6

modeling_prot2text.py CHANGED Viewed

@@ -123,9 +123,17 @@ class Prot2TextModel(PreTrainedModel):
     @torch.no_grad()
     def generate_protein_description(self,
-                                    protein_sequence=None,
-                                    tokenizer=None,
-                                    device='cpu'
                                      ):
         if self.config.esm and not self.config.rgcn and protein_sequence==None:
@@ -147,9 +155,23 @@ class Prot2TextModel(PreTrainedModel):
         inputs = {k: v.to(device=device, non_blocking=True) if hasattr(v, 'to') else v for k, v in inputs.items()}
         encoder_state = dict()
         encoder_state['hidden_states'] = self(**inputs, get_graph_emb=True, output_attentions=True)
-        generated = tokenizer.batch_decode(self.decoder.generate(input_ids=inputs['decoder_input_ids'], encoder_outputs=encoder_state, use_cache=True), skip_special_tokens=True)
-        return generated[0].replace('<|stop_token|>', '').replace('<|graph_token|>', '')
     @torch.no_grad()
     def generate(self,

     @torch.no_grad()
     def generate_protein_description(self,
+                                     protein_sequence=None,
+                                     tokenizer=None,
+                                     device='cpu',
+                                     streamer=None,
+                                     max_new_tokens=None,
+                                     do_sample=None,
+                                     top_p=None,
+                                     top_k=None,
+                                     temperature=None,
+                                     num_beams=1,
+                                     repetition_penalty=None
                                      ):
         if self.config.esm and not self.config.rgcn and protein_sequence==None:
         inputs = {k: v.to(device=device, non_blocking=True) if hasattr(v, 'to') else v for k, v in inputs.items()}
         encoder_state = dict()
         encoder_state['hidden_states'] = self(**inputs, get_graph_emb=True, output_attentions=True)
+        if streamer is None:
+            generated = tokenizer.batch_decode(self.decoder.generate(input_ids=inputs['decoder_input_ids'], encoder_outputs=encoder_state, use_cache=True), skip_special_tokens=True)
+            return generated[0].replace('<|stop_token|>', '').replace('<|graph_token|>', '')
+        else:
+            return self.decoder.generate(input_ids=inputs['decoder_input_ids'],
+                                         encoder_outputs=encoder_state,
+                                         use_cache=True,
+                                         streamer=streamer,
+                                         max_new_tokens=max_new_tokens,
+                                         do_sample=do_sample,
+                                         top_p=top_p,
+                                         top_k=top_k,
+                                         temperature=temperature,
+                                         num_beams=1,
+                                         repetition_penalty=repetition_penalty)
     @torch.no_grad()
     def generate(self,