Comments (5)
Hey @SunMarc, would you have some bandwidth to take a look at this ? :)
from transformers.
I found that the PPL issue is related to Llama3 or llama.cpp. It doesn't happen with TinyLlama. I'll create another issue to discuss if needed.
from transformers.
It's easy to support GGUF FP16. Since BF16 is not supported by NumPy, my current workaround is to convert BF16 to FP16 using PyTorch, but it's not ideal to rely on PyTorch at this step.
Reference: main...PenutChen:transformers:main
def load_dequant_gguf_tensor(shape, ggml_type, data):
if ggml_type == GGML_TYPES["F32"]:
values = data
elif ggml_type == GGML_TYPES["F16"]:
values = data
elif ggml_type == GGML_TYPES["BF16"]:
import torch
data_uint8 = data.view(np.uint8)
tensor_uint8 = torch.from_numpy(data_uint8)
values = tensor_uint8.view(torch.bfloat16).float().numpy()
Note that BF16 support requires modifying some code in gguf-py. Since the latest version of gguf-py from the llama.cpp repo doesn't work with the current HF integration (#31725), I modified the version from PyPI as follows:
class GGMLQuantizationType(IntEnum):
F32 = 0
F16 = 1
BF16 = 30
# ...
GGML_QUANT_SIZES = {
GGMLQuantizationType.F32: (1, 4),
GGMLQuantizationType.F16: (1, 2),
GGMLQuantizationType.BF16: (1, 2),
# ...
}
from transformers.
Hey @PenutChen, thanks for your research ! I think that we should just support FP16 first since supporting BF16 would require a new gguf release + transformers gguf integration is not compatible yet. LMK what you think ! If you have some time, would you like a open a PR ? Otherwise, I will do it !
from transformers.
@SunMarc Sure, I will do the necessary checks and open a PR! By the way, gguf-py on PyPI has not been updated for a long time. Most developers from llama.cpp seem to use gguf-py from the source. I think if we want to improve this integration, we should discuss it with the developers of llama.cpp.
from transformers.
Related Issues (20)
- Is there any way to update the parameters of embedding model? HOT 2
- A bug that may cause device inconsistency HOT 4
- Gemma 2 Inference with BF16 fails HOT 5
- cuda device is wrongly requested instead of xpu running pipeline(device_map="auto", max_memory": {0: 1.0e+10}) HOT 5
- Incorrect Whisper long-form decoding timestamps HOT 4
- Very different output depending on whether an attention mask is passed when using caching HOT 3
- `last_hidden_state` has a different shape than `hidden_states[-1]` in the output of `SeamlessM4Tv2SpeechEncoder` if adapter layers are present HOT 6
- [GroundingDino] - GroundingDinoProcessor kwargs is Broken HOT 2
- Flash Attention with Gemma 2 HOT 11
- FX tracer doen't work when requesting non-default input argument HOT 2
- Keep Tuple of past key values as an option HOT 9
- How to manually stop the LLM output? HOT 2
- Pipeline's "num_return_sequences" > greater than 1 causes a runtime error with Gemma-2-9B. HOT 6
- WavLM returns empty hidden states when loaded directly to GPU HOT 1
- "TypeError: Object of type device is not JSON serializable" when saving the model on TPU HOT 6
- Add Depth Anything v2 metric depth HOT 6
- `attention_mask` must be in the same device as model? HOT 1
- `Gemma2Model` not returning cache HOT 8
- the attention output from llama2 generate differs from other llama models HOT 3
- Whisper + Torch.Compile: torch._dynamo.exc.Unsupported: reconstruct: UserDefinedObjectVariable(EncoderDecoderCache) HOT 6
Recommend Projects
-
React
A declarative, efficient, and flexible JavaScript library for building user interfaces.
-
Vue.js
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
-
Typescript
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
-
TensorFlow
An Open Source Machine Learning Framework for Everyone
-
Django
The Web framework for perfectionists with deadlines.
-
Laravel
A PHP framework for web artisans
-
D3
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
-
Recommend Topics
-
javascript
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
-
web
Some thing interesting about web. New door for the world.
-
server
A server is a program made to process requests and deliver data to clients.
-
Machine learning
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
-
Visualization
Some thing interesting about visualization, use data art
-
Game
Some thing interesting about game, make everyone happy.
Recommend Org
-
Facebook
We are working to build community through open source technology. NB: members must have two-factor auth.
-
Microsoft
Open source projects and samples from Microsoft.
-
Google
Google ❤️ Open Source for everyone.
-
Alibaba
Alibaba Open Source for everyone
-
D3
Data-Driven Documents codes.
-
Tencent
China tencent open source team.
from transformers.