Comments (3)
I'm not certain this is the problem so it would be good to validate it. But fusing can cause precision issues. In low precision: c = a + b
can give very inexact results if a
and b
have very different magnitudes. For example if a
is big and b
is small then c = a + b = a
. In your case if the adapters have small values and the original weight matrix has large values, then fusing can wipe-out the adapters rendering the baseline model.
Now, I'm not sure that's happening. There's a couple things you could do to check.
- Inspect the magnitudes of the weights and adapters
- Try fusing and running the model in higher precision (e.g. fp32) just as a test that it works.
- Using a larger scale sometimes (but not always) helps here also.
Another option is to avoid fusing entirely. There may be a way to run unfused models with llama.cpp (see #816 (comment)). Or you could use MLX LM to run the fine-tuned model instead of using llama.cpp?
from mlx-examples.
Facing pretty much the same problem on my end too with a different model (Mistral). #849
from mlx-examples.
This makes sense :) Thank you. Will have to look into that :)
from mlx-examples.
Related Issues (20)
- [Model Request] Add support for IBM's Granite model HOT 2
- [Feature] Export Lora Adapters as GGML HOT 3
- Error when running inference on newly converted OpenELM MLX model, ValueError(f"Received parameters not in model: {extras}.") HOT 1
- LLMEvaluator : libc++abi: terminating due to uncaught exception of type std::invalid_argument: [matmul] Last dimension of first input with shape (1,916,2048) must match second to last dimension of second input with shape (256,32000)
- Unable to allocate memory
- Proposal: Add mypy to .pre-commit-config.yml HOT 2
- Struggling to convert models to MLX HOT 2
- mlx_lm stops generating HOT 1
- lora resume error HOT 2
- Error loading GGUF Mixtral 8x7B Q_8 model HOT 1
- iterate_batches in mlx_lm's Lora trainer is discarding the remainder dataset items (modulo batch size) HOT 1
- 01-ai/Yi-6B-Chat got IndexError: list assignment index out of range HOT 2
- [Feature Request] Finetuning Scripts for Whisper Models HOT 1
- Feature Request - Beam Search Decoder
- Discrepancies in generations from the fine tuned models after and before converting them into GGUF. The output generations go into an infinite loop. HOT 5
- NameError: name 'resume_adapter_file' is not defined HOT 1
- Received parameters not in model: {extras}. HOT 1
- support Gemma 2
- Model type deepseek_v2 not supported.
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 mlx-examples.