Instructions to use fischkas09/peptide-chat-qwen3 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use fischkas09/peptide-chat-qwen3 with PEFT:

from peft import PeftModel
from transformers import AutoModelForCausalLM

base_model = AutoModelForCausalLM.from_pretrained("Qwen/Qwen3-4B-Instruct-2507")
model = PeftModel.from_pretrained(base_model, "fischkas09/peptide-chat-qwen3")

Transformers

How to use fischkas09/peptide-chat-qwen3 with Transformers:

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

pipe = pipeline("text-generation", model="fischkas09/peptide-chat-qwen3")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("fischkas09/peptide-chat-qwen3")
model = AutoModelForCausalLM.from_pretrained("fischkas09/peptide-chat-qwen3")
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 fischkas09/peptide-chat-qwen3 with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "fischkas09/peptide-chat-qwen3"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "fischkas09/peptide-chat-qwen3",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/fischkas09/peptide-chat-qwen3

SGLang

How to use fischkas09/peptide-chat-qwen3 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 "fischkas09/peptide-chat-qwen3" \
    --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": "fischkas09/peptide-chat-qwen3",
		"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 "fischkas09/peptide-chat-qwen3" \
        --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": "fischkas09/peptide-chat-qwen3",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use fischkas09/peptide-chat-qwen3 with Docker Model Runner:
```
docker model run hf.co/fischkas09/peptide-chat-qwen3
```

See axolotl config

axolotl version: 0.13.0.dev0

base_model: Qwen/Qwen3-4B-Instruct-2507
# Automatically upload checkpoint and final model to HF
# hub_model_id: username/custom_model_name

plugins:
  - axolotl.integrations.cut_cross_entropy.CutCrossEntropyPlugin
strict: false

chat_template: qwen3
datasets:
  - path: fischkas09/TPDB_conversations
    type: chat_template
    split: train
    field_messages: messages
    message_property_mappings:
      role: role
      content: content
val_set_size: 0.15
output_dir: /workspace/axolotl/peptide_chat/model
dataset_prepared_path: /workspace/axolotl/peptide_chat/prepared_data

sequence_len: 4096
sample_packing: true
eval_sample_packing: true

load_in_4bit: true
adapter: qlora
lora_r: 8
lora_alpha: 16
lora_target_modules:
  - q_proj
  - k_proj
  - v_proj
  - o_proj
  - down_proj
  - up_proj
lora_mlp_kernel: true
lora_qkv_kernel: true
lora_o_kernel: true

use_wandb: true
wandb_name: qwen3-qlora-peptide-chat
wandb_project: peptide-chat
wandb_entity:
wandb_watch:
wandb_name:
wandb_log_model:

gradient_accumulation_steps: 8
micro_batch_size: 2
num_epochs: 3
optimizer: adamw_torch_4bit
lr_scheduler: cosine
learning_rate: 0.0001

bf16: auto
tf32: true

gradient_checkpointing: true
gradient_checkpointing_kwargs:
  use_reentrant: false
auto_resume_from_checkpoints: true
logging_steps: 1
flash_attention: true

warmup_ratio: 0.1
evals_per_epoch: 4
saves_per_epoch: 1
weight_decay: 0.0
special_tokens:

# save_first_step: true  # uncomment this to validate checkpoint saving works with your config

workspace/axolotl/peptide_chat/model

This model is a fine-tuned version of Qwen/Qwen3-4B-Instruct-2507 on the fischkas09/TPDB_conversations dataset. It achieves the following results on the evaluation set:

Loss: 4.3237
Ppl: 75.4694
Memory/max Active (gib): 3.33
Memory/max Allocated (gib): 3.33
Memory/device Reserved (gib): 6.3

Model description

More information needed

Intended uses & limitations

More information needed

Training and evaluation data

More information needed

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

learning_rate: 0.0001
train_batch_size: 2
eval_batch_size: 2
seed: 42
gradient_accumulation_steps: 8
total_train_batch_size: 16
optimizer: Use OptimizerNames.ADAMW_TORCH_4BIT with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
lr_scheduler_type: cosine
lr_scheduler_warmup_steps: 20
training_steps: 204

Training results

Training Loss	Epoch	Step	Validation Loss	Ppl	Active (gib)	Allocated (gib)	Reserved (gib)
No log	0	0	7.5089	1824.1217	3.31	3.31	5.3
6.1255	0.2468	17	5.8785	357.2578	3.33	3.33	6.3
4.8152	0.4936	34	4.8201	123.9785	3.33	3.33	6.3
4.6569	0.7405	51	4.6556	105.1744	3.33	3.33	6.31
4.5626	0.9873	68	4.5607	95.6505	3.33	3.33	6.3
4.4898	1.2323	85	4.4996	89.9799	3.33	3.33	6.3
4.4592	1.4791	102	4.4429	85.0226	3.33	3.33	6.3
4.4193	1.7260	119	4.3994	81.3985	3.33	3.33	6.3
4.3905	1.9728	136	4.3698	79.0246	3.33	3.33	6.3
4.2897	2.2178	153	4.3452	77.1110	3.33	3.33	6.3
4.318	2.4646	170	4.3318	76.0812	3.33	3.33	6.3
4.3452	2.7114	187	4.3248	75.5490	3.33	3.33	6.3
4.3629	2.9583	204	4.3237	75.4694	3.33	3.33	6.3

Framework versions

PEFT 0.18.1
Transformers 4.57.6
Pytorch 2.9.1+cu128
Datasets 4.5.0
Tokenizers 0.22.2

Downloads last month: -

Model tree for fischkas09/peptide-chat-qwen3

Base model

Qwen/Qwen3-4B-Instruct-2507

Adapter

(5501)

this model

fischkas09
/

peptide-chat-qwen3

workspace/axolotl/peptide_chat/model

Model description

Intended uses & limitations

Training and evaluation data

Training procedure

Training hyperparameters

Training results

Framework versions

Model tree for fischkas09/peptide-chat-qwen3

Dataset used to train fischkas09/peptide-chat-qwen3