Update README.md

README.md

@@ -43,10 +43,10 @@ datasets:
 - Quantized by [David Xue](https://www.linkedin.com/in/david-xue-uva/) from [Astronomer](https://astronomer.io)
 
 ## MUST READ: Very Important!! Note About Untrained Special Tokens in Llama 3 Base (Non-instruct) Models & Fine-tuning Llama 3 Base
+- **If you intend to fine-tune this model with any added tokens, or fine-tune for instruction following, please use the `untrained-special-tokens-fixed` branch/revision.**
 - Special tokens such as the ones used for instruct are undertrained in Llama 3 base models.
 - Credits: discovered by Daniel Han https://twitter.com/danielhanchen/status/1781395882925343058
 - ![image/png](https://cdn-uploads.huggingface.co/production/uploads/655ad0f8727df37c77a09cb9/1U2rRrx60p1pNeeAZw8Rd.png)
-- A patch function is under way, fine-tuning this model for instruction following may cause `NaN` graidents unless this problem is addressed.
 
 ## Important Note About Serving with vLLM & oobabooga/text-generation-webui
 - For loading this model onto vLLM, make sure all requests have `"stop_token_ids":[128001, 128009]` to temporarily address the non-stop generation issue.
@@ -72,9 +72,11 @@ The 8 bit GPTQ quant has minimum quality degradation from the original `bfloat16
 - This model is quantized by utilizing the AutoGPTQ library, following best practices noted by [GPTQ paper](https://arxiv.org/abs/2210.17323)
 - Quantization is calibrated and aligned with random samples from the specified dataset (wikitext for now) for minimum accuracy loss.
 
-| Branch | Bits | Group Size | Act Order | Damp % | GPTQ Dataset | Sequence Length | VRAM Size | ExLlama | Description |
+| Branch | Bits | Group Size | Act Order | Damp % | GPTQ Dataset | Sequence Length | VRAM Size | ExLlama | Special Tokens Fixed | Description |
 | ------ | ---- | -- | --------- | ------ | ------------ | ------- | ---- | ------- | ---- |
-| [main](https://huggingface.co/astronomer-io/Llama-3-8B-GPTQ-8-Bit/tree/main) | 8 | 32 | Yes | 0.1 | [wikitext](https://huggingface.co/datasets/wikitext/viewer/wikitext-2-v1/test) | 8192 | 9.74 GB | No | 8-bit, with Act Order and group size 32g. Minimum accuracy loss with decent VRAM usage reduction. | 
+| [main](https://huggingface.co/astronomer-io/Llama-3-8B-GPTQ-8-Bit/tree/main) | 8 | 32 | Yes | 0.1 | [wikitext](https://huggingface.co/datasets/wikitext/viewer/wikitext-2-v1/test) | 8192 | 9.74 GB | No | No | 8-bit, with Act Order and group size 32g. Minimum accuracy loss with decent VRAM usage reduction. | 
+| [untrained-special-tokens-fixed](https://huggingface.co/astronomer-io/Llama-3-8B-GPTQ-8-Bit/tree/untrained-special-tokens-fixed) | 8 | 32 | Yes | 0.1 | [wikitext](https://huggingface.co/datasets/wikitext/viewer/wikitext-2-v1/test) | 8192 | 9.74 GB | No | Yes | Same as the main branch. The special tokens that were untrained causing exploding graidents/NaN gradients have had their embedding values set to the average of trained tokens for each feature | 
+
 | More variants to come | TBD | TBD | TBD | TBD | TBD | TBD | TBD | TBD | May upload additional variants of GPTQ 8 bit models in the future using different parameters such as 128g group size and etc. |
 
 ## Serving this GPTQ model using vLLM