thinking-machines-lab/tinker
1
0
Fork 0
mirror of https://github.com/thinking-machines-lab/tinker.git synced 2026-04-26 17:13:11 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions

Sync contents

Browse source
This commit is contained in:
Daniel Xu 2025-12-15 01:00:20 +00:00
parent 5ad4282c96
commit e7a0d0ca2d
15 changed files with 91 additions and 31 deletions
Download patch file Download diff file Expand all files Collapse all files

2
docs/api/samplingclient.md
View file

@ -23,7 +23,7 @@ Args:
Example:
```python
sampling_client = service_client.create_sampling_client(base_model="Qwen/Qwen2.5-7B")
sampling_client = service_client.create_sampling_client(base_model="Qwen/Qwen3-8B")
prompt = types.ModelInput.from_ints(tokenizer.encode("The weather today is"))
params = types.SamplingParams(max_tokens=20, temperature=0.7)
future = sampling_client.sample(prompt=prompt, sampling_params=params, num_samples=1)

8
docs/api/serviceclient.md
View file

@ -74,7 +74,7 @@ def create_lora_training_client(
Create a TrainingClient for LoRA fine-tuning.
Args:
- `base_model`: Name of the base model to fine-tune (e.g., "Qwen/Qwen2.5-7B")
- `base_model`: Name of the base model to fine-tune (e.g., "Qwen/Qwen3-8B")
- `rank`: LoRA rank controlling the size of adaptation matrices (default 32)
- `seed`: Random seed for initialization. None means random seed.
- `train_mlp`: Whether to train MLP layers (default True)
@ -88,7 +88,7 @@ Returns:
Example:
```python
training_client = service_client.create_lora_training_client(
base_model="Qwen/Qwen2.5-7B",
base_model="Qwen/Qwen3-8B",
rank=16,
train_mlp=True,
train_attn=True
@ -203,7 +203,7 @@ Create a SamplingClient for text generation.
Args:
- `model_path`: Path to saved model weights (e.g., "tinker://run-id/weights/checkpoint-001")
- `base_model`: Name of base model to use (e.g., "Qwen/Qwen2.5-7B")
- `base_model`: Name of base model to use (e.g., "Qwen/Qwen3-8B")
- `retry_config`: Optional configuration for retrying failed requests
Returns:
@ -216,7 +216,7 @@ Example:
```python
# Use a base model
sampling_client = service_client.create_sampling_client(
base_model="Qwen/Qwen2.5-7B"
base_model="Qwen/Qwen3-8B"
)
# Or use saved weights

2
docs/api/trainingclient.md
View file

@ -21,7 +21,7 @@ Args:
Example:
```python
training_client = service_client.create_lora_training_client(base_model="Qwen/Qwen2.5-7B")
training_client = service_client.create_lora_training_client(base_model="Qwen/Qwen3-8B")
fwdbwd_future = training_client.forward_backward(training_data, "cross_entropy")
optim_future = training_client.optim_step(types.AdamParams(learning_rate=1e-4))
fwdbwd_result = fwdbwd_future.result() # Wait for gradients