> ## Documentation Index
> Fetch the complete documentation index at: https://docs.together.ai/llms.txt
> Use this file to discover all available pages before exploring further.
# LoRA vs. full fine-tuning
> Choose between LoRA and full fine-tuning, then tune LoRA's rank and target modules.
Together AI supports two fine-tuning implementations:
* **LoRA:** Trains a small set of adapter weights on top of the frozen base model.
* **Full fine-tuning:** Updates every weight in the base model.
LoRA is the default on Together AI, because it trains 0.1% to 1% of the parameters that full fine-tuning would, costs less, and produces a compact adapter rather than a full set of model weights.
Both [supervised fine-tuning](/docs/fine-tuning/supervised) and [preference fine-tuning](/docs/fine-tuning/preference-tuning) support LoRA and full fine-tuning.
## Choose a method
Use LoRA when:
* **You're starting a new fine-tune:** LoRA gets you a working model fastest and at the lowest cost.
* **You want to ship multiple adapters from the same base:** Adapters are small and can be swapped on a single hosted base model.
* **You're tuning style, format, or domain vocabulary:** These are the kinds of updates that LoRA handles best.
Use full fine-tuning when:
* **The base behavior needs a substantial change:** A model that doesn't know the task you're training for may need every weight updated, not just an adapter.
* **LoRA results plateau below your target:** Try increasing `lora_r` and `lora_alpha` first, and if quality still falls short, switch to full fine-tuning.
## Set the method on your job
The `lora` parameter defaults to `True`. Pass `lora=False` (or `--no-lora` on the CLI) to run a full fine-tune instead. Everything else about the job stays the same.
```python Python theme={null}
from together import Together
client = Together()
# LoRA (default) — lora=True is optional
job = client.fine_tuning.create(
training_file="",
model="meta-llama/Meta-Llama-3.1-8B-Instruct-Reference",
lora=True,
)
# Full fine-tuning
job = client.fine_tuning.create(
training_file="",
model="meta-llama/Meta-Llama-3.1-8B-Instruct-Reference",
lora=False,
)
```
```typescript TypeScript theme={null}
import Together from "together-ai";
const client = new Together();
// LoRA (default) — lora: true is optional
const loraJob = await client.fineTuning.create({
training_file: "",
model: "meta-llama/Meta-Llama-3.1-8B-Instruct-Reference",
lora: true,
});
// Full fine-tuning
const fullJob = await client.fineTuning.create({
training_file: "",
model: "meta-llama/Meta-Llama-3.1-8B-Instruct-Reference",
lora: false,
});
```
```bash CLI theme={null}
# LoRA (default)
tg fine-tuning create \
--training-file "" \
--model "meta-llama/Meta-Llama-3.1-8B-Instruct-Reference" \
--lora
# Full fine-tuning
tg fine-tuning create \
--training-file "" \
--model "meta-llama/Meta-Llama-3.1-8B-Instruct-Reference" \
--no-lora
```
## LoRA settings
For the parameters that tune LoRA itself (`lora_r`, `lora_alpha`, `lora_dropout`, `lora_trainable_modules`), see the [fine-tuning API reference](/reference/post-fine-tunes).
## Default target modules
When you don't set `lora_trainable_modules`, it defaults to `all-linear`, which applies LoRA to the modules listed for each model in the tables below. To customize, pass a comma-separated list of module names instead.
### Text models
| Model | Default target modules |
| ------------------------------------------------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
| `togethercomputer/llama-2-7b-chat` | `k_proj`, `up_proj`, `o_proj`, `q_proj`, `down_proj`, `v_proj`, `gate_proj` |
| `meta-llama/Meta-Llama-3-8B` | `k_proj`, `up_proj`, `o_proj`, `q_proj`, `down_proj`, `v_proj`, `gate_proj` |
| `meta-llama/Meta-Llama-3-8B-Instruct` | `k_proj`, `up_proj`, `o_proj`, `q_proj`, `down_proj`, `v_proj`, `gate_proj` |
| `meta-llama/Meta-Llama-3-70B-Instruct` | `k_proj`, `up_proj`, `o_proj`, `q_proj`, `down_proj`, `v_proj`, `gate_proj` |
| `meta-llama/Meta-Llama-3.1-8B-Reference` | `k_proj`, `up_proj`, `o_proj`, `q_proj`, `down_proj`, `v_proj`, `gate_proj` |
| `meta-llama/Meta-Llama-3.1-8B-Instruct-Reference` | `k_proj`, `up_proj`, `o_proj`, `q_proj`, `down_proj`, `v_proj`, `gate_proj` |
| `meta-llama/Meta-Llama-3.1-8B-131k-Reference` | `k_proj`, `up_proj`, `o_proj`, `q_proj`, `down_proj`, `v_proj`, `gate_proj` |
| `meta-llama/Meta-Llama-3.1-8B-131k-Instruct-Reference` | `k_proj`, `up_proj`, `o_proj`, `q_proj`, `down_proj`, `v_proj`, `gate_proj` |
| `meta-llama/Meta-Llama-3.1-70B-Reference` | `k_proj`, `up_proj`, `o_proj`, `q_proj`, `down_proj`, `v_proj`, `gate_proj` |
| `meta-llama/Meta-Llama-3.1-70B-Instruct-Reference` | `k_proj`, `up_proj`, `o_proj`, `q_proj`, `down_proj`, `v_proj`, `gate_proj` |
| `meta-llama/Meta-Llama-3.1-70B-32k-Reference` | `k_proj`, `up_proj`, `o_proj`, `q_proj`, `down_proj`, `v_proj`, `gate_proj` |
| `meta-llama/Meta-Llama-3.1-70B-32k-Instruct-Reference` | `k_proj`, `up_proj`, `o_proj`, `q_proj`, `down_proj`, `v_proj`, `gate_proj` |
| `meta-llama/Meta-Llama-3.1-70B-131k-Reference` | `k_proj`, `up_proj`, `o_proj`, `q_proj`, `down_proj`, `v_proj`, `gate_proj` |
| `meta-llama/Meta-Llama-3.1-70B-131k-Instruct-Reference` | `k_proj`, `up_proj`, `o_proj`, `q_proj`, `down_proj`, `v_proj`, `gate_proj` |
| `meta-llama/Llama-3.2-1B` | `k_proj`, `up_proj`, `o_proj`, `q_proj`, `down_proj`, `v_proj`, `gate_proj` |
| `meta-llama/Llama-3.2-1B-Instruct` | `k_proj`, `up_proj`, `o_proj`, `q_proj`, `down_proj`, `v_proj`, `gate_proj` |
| `meta-llama/Llama-3.2-3B` | `k_proj`, `up_proj`, `o_proj`, `q_proj`, `down_proj`, `v_proj`, `gate_proj` |
| `meta-llama/Llama-3.2-3B-Instruct` | `k_proj`, `up_proj`, `o_proj`, `q_proj`, `down_proj`, `v_proj`, `gate_proj` |
| `meta-llama/Llama-3.3-70B-Instruct-Reference` | `k_proj`, `up_proj`, `o_proj`, `q_proj`, `down_proj`, `v_proj`, `gate_proj` |
| `meta-llama/Llama-3.3-70B-32k-Instruct-Reference` | `k_proj`, `up_proj`, `o_proj`, `q_proj`, `down_proj`, `v_proj`, `gate_proj` |
| `meta-llama/Llama-3.3-70B-131k-Instruct-Reference` | `k_proj`, `up_proj`, `o_proj`, `q_proj`, `down_proj`, `v_proj`, `gate_proj` |
| `meta-llama/Llama-4-Scout-17B-16E` | `k_proj`, `o_proj`, `q_proj`, `v_proj`, `shared_expert.gate_proj`, `shared_expert.up_proj`, `shared_expert.down_proj`, `feed_forward.gate_proj`, `feed_forward.up_proj`, `feed_forward.down_proj` |
| `meta-llama/Llama-4-Scout-17B-16E-Instruct` | `k_proj`, `o_proj`, `q_proj`, `v_proj`, `shared_expert.gate_proj`, `shared_expert.up_proj`, `shared_expert.down_proj`, `feed_forward.gate_proj`, `feed_forward.up_proj`, `feed_forward.down_proj` |
| `meta-llama/Llama-4-Maverick-17B-128E` | `k_proj`, `o_proj`, `q_proj`, `v_proj` |
| `meta-llama/Llama-4-Maverick-17B-128E-Instruct` | `k_proj`, `o_proj`, `q_proj`, `v_proj` |
| `mistralai/Mistral-7B-v0.1` | `k_proj`, `up_proj`, `o_proj`, `q_proj`, `down_proj`, `v_proj`, `gate_proj` |
| `mistralai/Mistral-7B-Instruct-v0.2` | `k_proj`, `up_proj`, `o_proj`, `q_proj`, `down_proj`, `v_proj`, `gate_proj` |
| `mistralai/Mixtral-8x7B-v0.1` | `k_proj`, `o_proj`, `q_proj`, `v_proj` |
| `mistralai/Mixtral-8x7B-Instruct-v0.1` | `k_proj`, `o_proj`, `q_proj`, `v_proj` |
| `Qwen/Qwen2-1.5B` | `k_proj`, `up_proj`, `o_proj`, `q_proj`, `down_proj`, `v_proj`, `gate_proj` |
| `Qwen/Qwen2-1.5B-Instruct` | `k_proj`, `up_proj`, `o_proj`, `q_proj`, `down_proj`, `v_proj`, `gate_proj` |
| `Qwen/Qwen2-7B` | `k_proj`, `up_proj`, `o_proj`, `q_proj`, `down_proj`, `v_proj`, `gate_proj` |
| `Qwen/Qwen2-7B-Instruct` | `k_proj`, `up_proj`, `o_proj`, `q_proj`, `down_proj`, `v_proj`, `gate_proj` |
| `Qwen/Qwen2-72B` | `k_proj`, `up_proj`, `o_proj`, `q_proj`, `down_proj`, `v_proj`, `gate_proj` |
| `Qwen/Qwen2-72B-Instruct` | `k_proj`, `up_proj`, `o_proj`, `q_proj`, `down_proj`, `v_proj`, `gate_proj` |
| `Qwen/Qwen2.5-1.5B` | `k_proj`, `up_proj`, `o_proj`, `q_proj`, `down_proj`, `v_proj`, `gate_proj` |
| `Qwen/Qwen2.5-1.5B-Instruct` | `k_proj`, `up_proj`, `o_proj`, `q_proj`, `down_proj`, `v_proj`, `gate_proj` |
| `Qwen/Qwen2.5-3B` | `k_proj`, `up_proj`, `o_proj`, `q_proj`, `down_proj`, `v_proj`, `gate_proj` |
| `Qwen/Qwen2.5-3B-Instruct` | `k_proj`, `up_proj`, `o_proj`, `q_proj`, `down_proj`, `v_proj`, `gate_proj` |
| `Qwen/Qwen2.5-7B` | `k_proj`, `up_proj`, `o_proj`, `q_proj`, `down_proj`, `v_proj`, `gate_proj` |
| `Qwen/Qwen2.5-7B-Instruct` | `k_proj`, `up_proj`, `o_proj`, `q_proj`, `down_proj`, `v_proj`, `gate_proj` |
| `Qwen/Qwen2.5-14B` | `k_proj`, `up_proj`, `o_proj`, `q_proj`, `down_proj`, `v_proj`, `gate_proj` |
| `Qwen/Qwen2.5-14B-Instruct` | `k_proj`, `up_proj`, `o_proj`, `q_proj`, `down_proj`, `v_proj`, `gate_proj` |
| `Qwen/Qwen2.5-32B` | `k_proj`, `up_proj`, `o_proj`, `q_proj`, `down_proj`, `v_proj`, `gate_proj` |
| `Qwen/Qwen2.5-32B-Instruct` | `k_proj`, `up_proj`, `o_proj`, `q_proj`, `down_proj`, `v_proj`, `gate_proj` |
| `Qwen/Qwen2.5-72B` | `k_proj`, `up_proj`, `o_proj`, `q_proj`, `down_proj`, `v_proj`, `gate_proj` |
| `Qwen/Qwen2.5-72B-Instruct` | `k_proj`, `up_proj`, `o_proj`, `q_proj`, `down_proj`, `v_proj`, `gate_proj` |
| `Qwen/Qwen3-0.6B-Base` | `k_proj`, `up_proj`, `o_proj`, `q_proj`, `down_proj`, `v_proj`, `gate_proj` |
| `Qwen/Qwen3-0.6B` | `k_proj`, `up_proj`, `o_proj`, `q_proj`, `down_proj`, `v_proj`, `gate_proj` |
| `Qwen/Qwen3-1.7B-Base` | `k_proj`, `up_proj`, `o_proj`, `q_proj`, `down_proj`, `v_proj`, `gate_proj` |
| `Qwen/Qwen3-1.7B` | `k_proj`, `up_proj`, `o_proj`, `q_proj`, `down_proj`, `v_proj`, `gate_proj` |
| `Qwen/Qwen3-4B-Base` | `k_proj`, `up_proj`, `o_proj`, `q_proj`, `down_proj`, `v_proj`, `gate_proj` |
| `Qwen/Qwen3-4B` | `k_proj`, `up_proj`, `o_proj`, `q_proj`, `down_proj`, `v_proj`, `gate_proj` |
| `Qwen/Qwen3-8B-Base` | `k_proj`, `up_proj`, `o_proj`, `q_proj`, `down_proj`, `v_proj`, `gate_proj` |
| `Qwen/Qwen3-8B` | `k_proj`, `up_proj`, `o_proj`, `q_proj`, `down_proj`, `v_proj`, `gate_proj` |
| `Qwen/Qwen3-14B-Base` | `k_proj`, `up_proj`, `o_proj`, `q_proj`, `down_proj`, `v_proj`, `gate_proj` |
| `Qwen/Qwen3-14B` | `k_proj`, `up_proj`, `o_proj`, `q_proj`, `down_proj`, `v_proj`, `gate_proj` |
| `Qwen/Qwen3-32B` | `k_proj`, `up_proj`, `o_proj`, `q_proj`, `down_proj`, `v_proj`, `gate_proj` |
| `Qwen/Qwen3-30B-A3B-Base` | `k_proj`, `o_proj`, `q_proj`, `v_proj` |
| `Qwen/Qwen3-30B-A3B` | `k_proj`, `o_proj`, `q_proj`, `v_proj` |
| `Qwen/Qwen3-30B-A3B-Instruct-2507` | `k_proj`, `o_proj`, `q_proj`, `v_proj` |
| `Qwen/Qwen3-235B-A22B` | `k_proj`, `o_proj`, `q_proj`, `v_proj` |
| `Qwen/Qwen3-235B-A22B-Instruct-2507` | `k_proj`, `o_proj`, `q_proj`, `v_proj` |
| `Qwen/Qwen3-Coder-30B-A3B-Instruct` | `k_proj`, `o_proj`, `q_proj`, `v_proj` |
| `Qwen/Qwen3-Coder-480B-A35B-Instruct` | `k_proj`, `o_proj`, `q_proj`, `v_proj` |
| `Qwen/Qwen3-Next-80B-A3B-Instruct` | `k_proj`, `o_proj`, `q_proj`, `v_proj` |
| `Qwen/Qwen3-Next-80B-A3B-Thinking` | `k_proj`, `o_proj`, `q_proj`, `v_proj` |
| `google/gemma-3-270m` | `k_proj`, `up_proj`, `o_proj`, `q_proj`, `down_proj`, `v_proj`, `gate_proj` |
| `google/gemma-3-270m-it` | `k_proj`, `up_proj`, `o_proj`, `q_proj`, `down_proj`, `v_proj`, `gate_proj` |
| `google/gemma-3-1b-it` | `k_proj`, `up_proj`, `o_proj`, `q_proj`, `down_proj`, `v_proj`, `gate_proj` |
| `google/gemma-3-1b-pt` | `k_proj`, `up_proj`, `o_proj`, `q_proj`, `down_proj`, `v_proj`, `gate_proj` |
| `google/gemma-3-4b-it` | `k_proj`, `up_proj`, `o_proj`, `q_proj`, `down_proj`, `v_proj`, `gate_proj` |
| `google/gemma-3-4b-pt` | `k_proj`, `up_proj`, `o_proj`, `q_proj`, `down_proj`, `v_proj`, `gate_proj` |
| `google/gemma-3-12b-it` | `k_proj`, `up_proj`, `o_proj`, `q_proj`, `down_proj`, `v_proj`, `gate_proj` |
| `google/gemma-3-12b-pt` | `k_proj`, `up_proj`, `o_proj`, `q_proj`, `down_proj`, `v_proj`, `gate_proj` |
| `google/gemma-3-27b-it` | `k_proj`, `up_proj`, `o_proj`, `q_proj`, `down_proj`, `v_proj`, `gate_proj` |
| `google/gemma-3-27b-pt` | `k_proj`, `up_proj`, `o_proj`, `q_proj`, `down_proj`, `v_proj`, `gate_proj` |
| `deepseek-ai/DeepSeek-R1-Distill-Qwen-1.5B` | `k_proj`, `up_proj`, `o_proj`, `q_proj`, `down_proj`, `v_proj`, `gate_proj` |
| `deepseek-ai/DeepSeek-R1-Distill-Qwen-14B` | `k_proj`, `up_proj`, `o_proj`, `q_proj`, `down_proj`, `v_proj`, `gate_proj` |
| `deepseek-ai/DeepSeek-R1-Distill-Llama-70B` | `k_proj`, `up_proj`, `o_proj`, `q_proj`, `down_proj`, `v_proj`, `gate_proj` |
| `deepseek-ai/DeepSeek-R1-Distill-Llama-70B-32k` | `k_proj`, `up_proj`, `o_proj`, `q_proj`, `down_proj`, `v_proj`, `gate_proj` |
| `deepseek-ai/DeepSeek-R1-Distill-Llama-70B-131k` | `k_proj`, `up_proj`, `o_proj`, `q_proj`, `down_proj`, `v_proj`, `gate_proj` |
| `deepseek-ai/DeepSeek-V3` | `q_a_proj`, `q_b_proj`, `kv_a_proj_with_mqa`, `kv_b_proj`, `mlp.gate_proj`, `mlp.up_proj`, `mlp.down_proj` |
| `deepseek-ai/DeepSeek-R1` | `q_a_proj`, `q_b_proj`, `kv_a_proj_with_mqa`, `kv_b_proj`, `mlp.gate_proj`, `mlp.up_proj`, `mlp.down_proj` |
| `deepseek-ai/DeepSeek-V3-Base` | `q_a_proj`, `q_b_proj`, `kv_a_proj_with_mqa`, `kv_b_proj`, `mlp.gate_proj`, `mlp.up_proj`, `mlp.down_proj` |
| `deepseek-ai/DeepSeek-V3-0324` | `q_a_proj`, `q_b_proj`, `kv_a_proj_with_mqa`, `kv_b_proj`, `mlp.gate_proj`, `mlp.up_proj`, `mlp.down_proj` |
| `deepseek-ai/DeepSeek-R1-0528` | `q_a_proj`, `q_b_proj`, `kv_a_proj_with_mqa`, `kv_b_proj`, `mlp.gate_proj`, `mlp.up_proj`, `mlp.down_proj` |
| `deepseek-ai/DeepSeek-V3.1-Base` | `q_a_proj`, `q_b_proj`, `kv_a_proj_with_mqa`, `kv_b_proj`, `mlp.gate_proj`, `mlp.up_proj`, `mlp.down_proj` |
| `deepseek-ai/DeepSeek-V3.1` | `q_a_proj`, `q_b_proj`, `kv_a_proj_with_mqa`, `kv_b_proj`, `mlp.gate_proj`, `mlp.up_proj`, `mlp.down_proj` |
| `moonshotai/Kimi-K2-Instruct` | `q_a_proj`, `q_b_proj`, `kv_a_proj_with_mqa`, `kv_b_proj`, `mlp.gate_proj`, `mlp.up_proj`, `mlp.down_proj` |
| `moonshotai/Kimi-K2-Thinking` | `q_a_proj`, `q_b_proj`, `kv_a_proj_with_mqa`, `kv_b_proj`, `mlp.gate_proj`, `mlp.up_proj`, `mlp.down_proj` |
| `moonshotai/Kimi-K2-Base` | `q_a_proj`, `q_b_proj`, `kv_a_proj_with_mqa`, `kv_b_proj`, `mlp.gate_proj`, `mlp.up_proj`, `mlp.down_proj` |
| `moonshotai/Kimi-K2-Instruct-0905` | `q_a_proj`, `q_b_proj`, `kv_a_proj_with_mqa`, `kv_b_proj`, `mlp.gate_proj`, `mlp.up_proj`, `mlp.down_proj` |
| `moonshotai/Kimi-K2.5` | `q_a_proj`, `q_b_proj`, `kv_a_proj_with_mqa`, `kv_b_proj`, `mlp.gate_proj`, `mlp.up_proj`, `mlp.down_proj` |
| `openai/gpt-oss-20b` | `k_proj`, `o_proj`, `q_proj`, `v_proj` |
| `openai/gpt-oss-120b` | `k_proj`, `o_proj`, `q_proj`, `v_proj` |
| `zai-org/GLM-4.6` | `k_proj`, `o_proj`, `q_proj`, `v_proj` |
| `zai-org/GLM-4.7` | `k_proj`, `o_proj`, `q_proj`, `v_proj` |
### Multimodal models
| Model | Default target modules |
| --------------------------------------------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
| `meta-llama/Llama-4-Scout-17B-16E-Instruct-VLM` | `k_proj`, `o_proj`, `q_proj`, `v_proj`, `shared_expert.gate_proj`, `shared_expert.up_proj`, `shared_expert.down_proj`, `feed_forward.gate_proj`, `feed_forward.up_proj`, `feed_forward.down_proj` |
| `meta-llama/Llama-4-Maverick-17B-128E-Instruct-VLM` | `k_proj`, `o_proj`, `q_proj`, `v_proj` |
| `Qwen/Qwen3-VL-8B-Instruct` | `k_proj`, `up_proj`, `o_proj`, `q_proj`, `down_proj`, `v_proj`, `gate_proj` |
| `Qwen/Qwen3-VL-32B-Instruct` | `k_proj`, `up_proj`, `o_proj`, `q_proj`, `down_proj`, `v_proj`, `gate_proj` |
| `Qwen/Qwen3-VL-30B-A3B-Instruct` | `k_proj`, `o_proj`, `q_proj`, `v_proj` |
| `Qwen/Qwen3-VL-235B-A22B-Instruct` | `k_proj`, `o_proj`, `q_proj`, `v_proj` |
| `Qwen/Qwen3.5-35B-A3B` | `k_proj`, `o_proj`, `q_proj`, `v_proj` |
| `Qwen/Qwen3.5-35B-A3B-Base` | `k_proj`, `o_proj`, `q_proj`, `v_proj` |
| `Qwen/Qwen3.5-122B-A10B` | `k_proj`, `o_proj`, `q_proj`, `v_proj` |
| `Qwen/Qwen3.5-397B-A17B` | `k_proj`, `o_proj`, `q_proj`, `v_proj` |
| `google/gemma-3-4b-it-VLM` | `k_proj`, `up_proj`, `o_proj`, `q_proj`, `down_proj`, `v_proj`, `gate_proj` |
| `google/gemma-3-12b-it-VLM` | `k_proj`, `up_proj`, `o_proj`, `q_proj`, `down_proj`, `v_proj`, `gate_proj` |
| `google/gemma-3-27b-it-VLM` | `k_proj`, `up_proj`, `o_proj`, `q_proj`, `down_proj`, `v_proj`, `gate_proj` |
| `google/gemma-4-31B-it-VLM` | `k_proj`, `up_proj`, `o_proj`, `q_proj`, `down_proj`, `v_proj`, `gate_proj` |
## What to expect from full fine-tuning
* **Supported models:** Full fine-tuning is available for a subset of the models that support LoRA. Large mixture-of-experts models, long-context variants, and some vision-language models are LoRA-only. See [supported models](/docs/fine-tuning/supported-models) for the per-model breakdown.
* **Smaller batch sizes:** Because full fine-tuning updates every weight, it carries a larger memory footprint, so the maximum batch size for a given model is generally smaller than the LoRA equivalent.
* **Higher cost:** Full fine-tuning trains every parameter rather than the 0.1% to 1% a LoRA job touches, so it consumes more compute and costs more. See [pricing](/docs/fine-tuning/pricing) for details.
## Serve your model
How you deploy depends on the method:
* **LoRA:** After the job completes, deploy the merged model on a dedicated endpoint, or upload just the adapter and serve it on a dedicated endpoint that runs it on top of the base model. See [deployment](/docs/fine-tuning/deployment) and [adapter upload](/docs/dedicated-endpoints/adapter).
* **Full fine-tuning:** The job produces a complete model rather than a compact adapter. Deploy it on a dedicated endpoint, or download the weights for local use. See [deployment](/docs/fine-tuning/deployment).