Spaces:
Running
on
Zero
Running
on
Zero
Switched model to OpenELM-3B-Instruct
Browse files
README.md
CHANGED
|
@@ -1,5 +1,5 @@
|
|
| 1 |
---
|
| 2 |
-
title: Apple OpenELM-3B
|
| 3 |
emoji: π
|
| 4 |
colorFrom: green
|
| 5 |
colorTo: red
|
|
@@ -15,7 +15,7 @@ suggested_hardware: t4-small
|
|
| 15 |
|
| 16 |
OpenELM was introduced in [this paper](https://arxiv.org/abs/2404.14619v1).
|
| 17 |
|
| 18 |
-
This Space demonstrates [OpenELM-3B](apple/OpenELM-3B) from Apple. Please, check the original model card for details.
|
| 19 |
You can see the other models of the OpenELM family [here](https://huggingface.co/apple/OpenELM)
|
| 20 |
|
| 21 |
# The following Information was taken "as is" from original model card
|
|
|
|
| 1 |
---
|
| 2 |
+
title: Apple OpenELM-3B-Instruct
|
| 3 |
emoji: π
|
| 4 |
colorFrom: green
|
| 5 |
colorTo: red
|
|
|
|
| 15 |
|
| 16 |
OpenELM was introduced in [this paper](https://arxiv.org/abs/2404.14619v1).
|
| 17 |
|
| 18 |
+
This Space demonstrates [OpenELM-3B-Instruct](https://huggingface.co/apple/OpenELM-3B-Instruct) from Apple. Please, check the original model card for details.
|
| 19 |
You can see the other models of the OpenELM family [here](https://huggingface.co/apple/OpenELM)
|
| 20 |
|
| 21 |
# The following Information was taken "as is" from original model card
|
app.py
CHANGED
|
@@ -12,12 +12,12 @@ DEFAULT_MAX_NEW_TOKENS = 256
|
|
| 12 |
MAX_INPUT_TOKEN_LENGTH = 512
|
| 13 |
|
| 14 |
DESCRIPTION = """\
|
| 15 |
-
# OpenELM-3B
|
| 16 |
|
| 17 |
-
This Space demonstrates [OpenELM-3B](https://huggingface.co/apple/OpenELM-3B) by Apple. Please, check the original model card for details.
|
| 18 |
You can see the other models of the OpenELM family [here](https://huggingface.co/apple/OpenELM)
|
| 19 |
The following Colab notebooks are available:
|
| 20 |
-
* [OpenELM-3B (GPU)](https://gist.github.com/Norod/4f11bb36bea5c548d18f10f9d7ec09b0)
|
| 21 |
* [OpenELM-270M (CPU)](https://gist.github.com/Norod/5a311a8e0a774b5c35919913545b7af4)
|
| 22 |
|
| 23 |
You might also be interested in checking out Apple's [CoreNet Github page](https://github.com/apple/corenet?tab=readme-ov-file).
|
|
@@ -33,8 +33,8 @@ LICENSE = """
|
|
| 33 |
<p/>
|
| 34 |
|
| 35 |
---
|
| 36 |
-
As a derivative work of [OpenELM-3B](https://huggingface.co/apple/OpenELM-3B) by Apple,
|
| 37 |
-
this demo is governed by the original [license](https://huggingface.co/apple/OpenELM-3B/blob/main/LICENSE).
|
| 38 |
"""
|
| 39 |
|
| 40 |
if not torch.cuda.is_available():
|
|
@@ -42,7 +42,7 @@ if not torch.cuda.is_available():
|
|
| 42 |
|
| 43 |
|
| 44 |
if torch.cuda.is_available():
|
| 45 |
-
model_id = "apple/OpenELM-3B"
|
| 46 |
model = AutoModelForCausalLM.from_pretrained(model_id, device_map="auto", trust_remote_code=True, low_cpu_mem_usage=True)
|
| 47 |
tokenizer_id = "meta-llama/Llama-2-7b-hf"
|
| 48 |
tokenizer = AutoTokenizer.from_pretrained(tokenizer_id)
|
|
|
|
| 12 |
MAX_INPUT_TOKEN_LENGTH = 512
|
| 13 |
|
| 14 |
DESCRIPTION = """\
|
| 15 |
+
# OpenELM-3B-Instruct
|
| 16 |
|
| 17 |
+
This Space demonstrates [OpenELM-3B-Instruct](https://huggingface.co/apple/OpenELM-3B-Instruct) by Apple. Please, check the original model card for details.
|
| 18 |
You can see the other models of the OpenELM family [here](https://huggingface.co/apple/OpenELM)
|
| 19 |
The following Colab notebooks are available:
|
| 20 |
+
* [OpenELM-3B-Instruct (GPU)](https://gist.github.com/Norod/4f11bb36bea5c548d18f10f9d7ec09b0)
|
| 21 |
* [OpenELM-270M (CPU)](https://gist.github.com/Norod/5a311a8e0a774b5c35919913545b7af4)
|
| 22 |
|
| 23 |
You might also be interested in checking out Apple's [CoreNet Github page](https://github.com/apple/corenet?tab=readme-ov-file).
|
|
|
|
| 33 |
<p/>
|
| 34 |
|
| 35 |
---
|
| 36 |
+
As a derivative work of [OpenELM-3B-Instruct](https://huggingface.co/apple/OpenELM-3B-Instruct) by Apple,
|
| 37 |
+
this demo is governed by the original [license](https://huggingface.co/apple/OpenELM-3B-Instruct/blob/main/LICENSE).
|
| 38 |
"""
|
| 39 |
|
| 40 |
if not torch.cuda.is_available():
|
|
|
|
| 42 |
|
| 43 |
|
| 44 |
if torch.cuda.is_available():
|
| 45 |
+
model_id = "apple/OpenELM-3B-Instruct"
|
| 46 |
model = AutoModelForCausalLM.from_pretrained(model_id, device_map="auto", trust_remote_code=True, low_cpu_mem_usage=True)
|
| 47 |
tokenizer_id = "meta-llama/Llama-2-7b-hf"
|
| 48 |
tokenizer = AutoTokenizer.from_pretrained(tokenizer_id)
|