michaelfeil
commited on
Commit
•
909ddf7
1
Parent(s):
eadf5aa
Update README.md (#13)
Browse files- Update README.md (0eac4d2059c15988de0624a0aa2024dea3fd75f2)
README.md
CHANGED
@@ -50,6 +50,7 @@ For training data, we generate long contexts by augmenting [SlimPajama](https://
|
|
50 |
| GPU Type | NVIDIA L40S | NVIDIA L40S | NVIDIA L40S | NVIDIA L40S |
|
51 |
| Minutes to Train (Wall)| 202 | 555 | 61 | 87 |
|
52 |
|
|
|
53 |
**Evaluation:**
|
54 |
|
55 |
![image/png](https://cdn-uploads.huggingface.co/production/uploads/6585dc9be92bc5f258156bd6/mWxIGZNi3ejlmeIDWafKu.png)
|
@@ -75,8 +76,11 @@ HAYSTACK3:
|
|
75 |
All boxes not pictured for Haystack 1 and 3 are 100% accurate. Haystacks 1,2 and 3 are further detailed in this [blog post](https://gradient.ai/blog/the-haystack-matters-for-niah-evals).
|
76 |
|
77 |
**Quants:**
|
78 |
-
- [GGUF](https://huggingface.co/crusoeai/Llama-3-8B-Instruct-1048k-GGUF)
|
79 |
- [MLX-4bit](https://huggingface.co/mlx-community/Llama-3-8B-Instruct-1048k-4bit)
|
|
|
|
|
|
|
80 |
|
81 |
## The Gradient AI Team
|
82 |
|
|
|
50 |
| GPU Type | NVIDIA L40S | NVIDIA L40S | NVIDIA L40S | NVIDIA L40S |
|
51 |
| Minutes to Train (Wall)| 202 | 555 | 61 | 87 |
|
52 |
|
53 |
+
|
54 |
**Evaluation:**
|
55 |
|
56 |
![image/png](https://cdn-uploads.huggingface.co/production/uploads/6585dc9be92bc5f258156bd6/mWxIGZNi3ejlmeIDWafKu.png)
|
|
|
76 |
All boxes not pictured for Haystack 1 and 3 are 100% accurate. Haystacks 1,2 and 3 are further detailed in this [blog post](https://gradient.ai/blog/the-haystack-matters-for-niah-evals).
|
77 |
|
78 |
**Quants:**
|
79 |
+
- [GGUF by Crusoe](https://huggingface.co/crusoeai/Llama-3-8B-Instruct-1048k-GGUF). Note that you need to add 128009 as [special token with llama.cpp](https://huggingface.co/gradientai/Llama-3-8B-Instruct-262k/discussions/13).
|
80 |
- [MLX-4bit](https://huggingface.co/mlx-community/Llama-3-8B-Instruct-1048k-4bit)
|
81 |
+
- [Ollama](https://ollama.com/library/llama3-gradient)
|
82 |
+
- vLLM docker image, recommended to load via `--max-model-len 32768`
|
83 |
+
- If you are interested in a hosted version, drop us a mail below.
|
84 |
|
85 |
## The Gradient AI Team
|
86 |
|