Compare model versions

Diff

Before

<details>
<summary>Sentence Transformers</summary>

```python
from sentence_transformers import SentenceTransformer

model = SentenceTransformer(
    "ai-sage/Giga-Embeddings-instruct-3B-0826",
    trust_remote_code=True,   # needed for the bidirectional modeling code
)

instruction = "Given a query, retrieve relevant passages"
queries = [f"Instruct: {instruction}\nQuery: Где столица России?"]
documents = ["Москва — столица Российской Федерации.",
             "Париж — столица Франции."]

q_emb = model.encode(queries, normalize_embeddings=True)
d_emb = model.encode(documents, normalize_embeddings=True)
print(model.similarity(q_emb, d_emb))
```
</details>


<details>
<summary>Transformers (manual mean pooling)</summary>

```python
import torch
import torch.nn.functional as F
from transformers import AutoModel, AutoTokenizer

path = "ai-sage/Giga-Embeddings-instruct-3B-0826"
tok = AutoTokenizer.from_pretrained(path, trust_remote_code=True)
model = AutoModel.from_pretrained(path, trust_remote_code=True,
                                  dtype=torch.bfloat16).cuda().eval()

def encode(texts):
    enc = tok(texts, return_tensors="pt", padding=True, truncation=True, max_length=512)
    enc = {k: v.cuda() for k, v in enc.items()}
    with torch.no_grad():
        hidden = model(**enc).last_hidden_state
    mask = enc["attention_mask"].unsqueeze(-1).to(hidden.dtype)
    emb = (hidden * mask).sum(1) / mask.sum(1).clamp(min=1e-6)   # mean pool
    return F.normalize(emb, dim=-1)                              # L2 normalize

instr = "Given a query, retrieve relevant passages"
q = encode([f"Instruct: {instr}\nQuery: Где столица России?"])
d = encode(["Москва — столица Российской Федерации.", "Париж — столица Франции."])
print((q @ d.T).cpu())
```

</details>

<details>
<summary>vLLM</summary>

vLLM serves this model as an embedding model using its native Qwen3 implementation. Bidirectional attention is enabled with `is_causal=false`; no custom code is required on the vLLM side.

```python
from vllm import LLM
from vllm.config import PoolerConfig

llm = LLM(
    model="ai-sage/Giga-Embeddings-instruct-3B-0826",
    runner="pooling",
    convert="embed",
    hf_overrides={"is_causal": False, "architectures": ["Qwen3ForCausalLM"]},
    pooler_config=PoolerConfig(pooling_type="MEAN", use_activation=True),
    trust_remote_code=True,
)

instr = "Given a query, retrieve relevant passages"
outs = llm.encode([f"Instruct: {instr}\nQuery: Где столица России?",
                   "Москва — столица Российской Федерации."],
                  pooling_task="embed")
embs = [o.outputs.data for o in outs]
```

Or via the OpenAI-compatible server:

26 more lines not shown.

After

<details>
<summary>Sentence Transformers</summary>

```python
from sentence_transformers import SentenceTransformer

model = SentenceTransformer(
    "ai-sage/Giga-Embeddings-instruct-3B-0826",
    trust_remote_code=True,   # needed for the bidirectional modeling code
)

instruction = "Given a query, retrieve relevant passages"
queries = [f"Instruct: {instruction}\nQuery: Где столица России?"]
documents = ["Москва — столица Российской Федерации.",
             "Париж — столица Франции."]

q_emb = model.encode(queries, normalize_embeddings=True)
d_emb = model.encode(documents, normalize_embeddings=True)
print(model.similarity(q_emb, d_emb))
```
</details>


<details>
<summary>Transformers (manual mean pooling)</summary>

```python
import torch
import torch.nn.functional as F
from transformers import AutoModel, AutoTokenizer

path = "ai-sage/Giga-Embeddings-instruct-3B-0826"
tok = AutoTokenizer.from_pretrained(path, trust_remote_code=True)
model = AutoModel.from_pretrained(path, trust_remote_code=True,
                                  dtype=torch.bfloat16).cuda().eval()

def encode(texts):
    enc = tok(texts, return_tensors="pt", padding=True, truncation=True, max_length=512)
    enc = {k: v.cuda() for k, v in enc.items()}
    with torch.no_grad():
        hidden = model(**enc).last_hidden_state
    mask = enc["attention_mask"].unsqueeze(-1).to(hidden.dtype)
    emb = (hidden * mask).sum(1) / mask.sum(1).clamp(min=1e-6)   # mean pool
    return F.normalize(emb, dim=-1)                              # L2 normalize

instr = "Given a query, retrieve relevant passages"
q = encode([f"Instruct: {instr}\nQuery: Где столица России?"])
d = encode(["Москва — столица Российской Федерации.", "Париж — столица Франции."])
print((q @ d.T).cpu())
```

</details>

<details>
<summary>vLLM</summary>

vLLM serves this model as an embedding model using its native Qwen3 implementation. Bidirectional attention is enabled with `is_causal=false`; no custom code is required on the vLLM side.

Install the latest vLLM release from PyPI:

```bash
pip install --upgrade vllm
```

```python
from vllm import LLM
from vllm.config import PoolerConfig

llm = LLM(
    model="ai-sage/Giga-Embeddings-instruct-3B-0826",
    runner="pooling",
    convert="embed",
    hf_overrides={"is_causal": False, "architectures": ["Qwen3ForCausalLM"]},
    pooler_config=PoolerConfig(pooling_type="MEAN", use_activation=True),
    trust_remote_code=True,
)

instr = "Given a query, retrieve relevant passages"
outs = llm.encode([f"Instruct: {instr}\nQuery: Где столица России?",
                   "Москва — столица Российской Федерации."],

32 more lines not shown.

Diff

Before

```bash
# ms-swift from main — the current PyPI release does not yet contain the
# SentenceTransformer full-parameter save fix.
pip install "git+https://github.com/modelscope/ms-swift.git"

# sentence-transformers pinned to 5.3.0.
pip install "sentence-transformers==5.3.0"
```

torch and transformers are pulled in automatically. A CUDA GPU is required for training.

For **full-parameter finetuning of the 10B MoE model** (multi-GPU, DeepSpeed ZeRO-3),
also install DeepSpeed:

```bash
pip install deepspeed
```

---

After

```bash
# Install the latest ms-swift release from PyPI.
pip install --upgrade ms-swift

# sentence-transformers pinned to 5.3.0.
pip install "sentence-transformers==5.3.0"
```

torch and transformers are pulled in automatically. A CUDA GPU is required for training.

For **full-parameter finetuning of the 10B MoE model** (multi-GPU, DeepSpeed ZeRO-3),
also install DeepSpeed:

```bash
pip install deepspeed
```

---

Diff

Before

- **`AttributeError: 'NoneType' object has no attribute 'items'` during save** — you're
  on the PyPI release of ms-swift; install from `main` (see Requirements).
- **Reloaded model gives base-model quality / "missing keys" on load after full
  finetuning** — your `sentence-transformers` is newer than 5.3; pin `==5.3.0`.
- **Model downloads from ModelScope instead of HuggingFace (or is not found)** — add
  `--use_hf true`.

</details>

[Paper](https://arxiv.org/abs/2608.23806)

After

- **`AttributeError: 'NoneType' object has no attribute 'items'` during save** — upgrade
  ms-swift with `pip install --upgrade ms-swift` to get the SentenceTransformer
  full-parameter save fix (see Requirements).
- **Reloaded model gives base-model quality / "missing keys" on load after full
  finetuning** — your `sentence-transformers` is newer than 5.3; pin `==5.3.0`.
- **Model downloads from ModelScope instead of HuggingFace (or is not found)** — add
  `--use_hf true`.

</details>

[Paper](https://arxiv.org/abs/2608.23806)