Update README.md
Browse files
README.md
CHANGED
@@ -3,28 +3,25 @@ license: cc-by-sa-4.0
|
|
3 |
inference: false
|
4 |
---
|
5 |
|
6 |
-
# SLIM-
|
7 |
|
8 |
<!-- Provide a quick summary of what the model is/does. -->
|
9 |
|
10 |
-
**slim-
|
11 |
|
12 |
-
|
13 |
-
|
14 |
-
This 'combo' model is designed to illustrate the potential power of using function calls on small, specialized models to enable a single model architecture to combine the capabilities of what were traditionally two separate model architectures on an encoder.
|
15 |
-
|
16 |
-
The intent of SLIMs is to forge a middle-ground between traditional encoder-based classifiers and open-ended API-based LLMs, providing an intuitive, flexible natural language response, without complex prompting, and with improved generalization and ability to fine-tune to a specific domain use case.
|
17 |
|
|
|
18 |
|
19 |
This model is fine-tuned on top of [**llmware/bling-stable-lm-3b-4e1t-v0**](https://huggingface.co/llmware/bling-stable-lm-3b-4e1t-v0), which in turn, is a fine-tune of stabilityai/stablelm-3b-4elt.
|
20 |
|
21 |
-
Each slim model has a 'quantized tool' version, e.g., [**'slim-
|
22 |
|
23 |
|
24 |
## Prompt format:
|
25 |
|
26 |
-
`function = "
|
27 |
-
`params = "
|
28 |
`prompt = "<human> " + {text} + "\n" + `
|
29 |
`"<{function}> " + {params} + "</{function}>" + "\n<bot>:"`
|
30 |
|
@@ -32,11 +29,11 @@ Each slim model has a 'quantized tool' version, e.g., [**'slim-sa-ner-3b-tool'*
|
|
32 |
<details>
|
33 |
<summary>Transformers Script </summary>
|
34 |
|
35 |
-
model = AutoModelForCausalLM.from_pretrained("llmware/slim-
|
36 |
-
tokenizer = AutoTokenizer.from_pretrained("llmware/slim-
|
37 |
|
38 |
-
function = "
|
39 |
-
params = "
|
40 |
|
41 |
text = "Tesla stock declined yesterday 8% in premarket trading after a poorly-received event in San Francisco yesterday, in which the company indicated a likely shortfall in revenue."
|
42 |
|
@@ -75,8 +72,8 @@ Each slim model has a 'quantized tool' version, e.g., [**'slim-sa-ner-3b-tool'*
|
|
75 |
<summary>Using as Function Call in LLMWare</summary>
|
76 |
|
77 |
from llmware.models import ModelCatalog
|
78 |
-
slim_model = ModelCatalog().load_model("llmware/slim-
|
79 |
-
response = slim_model.function_call(text,params=["
|
80 |
|
81 |
print("llmware - llm_response: ", response)
|
82 |
|
|
|
3 |
inference: false
|
4 |
---
|
5 |
|
6 |
+
# SLIM-SUMMARY
|
7 |
|
8 |
<!-- Provide a quick summary of what the model is/does. -->
|
9 |
|
10 |
+
**slim-summary** is a small, specialized model finetuned for summarize function-calls, generating output consisting of a python dictionary with a "summary" key, and a value that consists of a list of distinct summary points.
|
11 |
|
12 |
+
As an experimental feature in the model, there is an optional list size that can be passed with the parameters in invoking the model to guide the model to a specific number of response elements.
|
|
|
|
|
|
|
|
|
13 |
|
14 |
+
`{'summary': ['point1', 'point2', 'point3']}`
|
15 |
|
16 |
This model is fine-tuned on top of [**llmware/bling-stable-lm-3b-4e1t-v0**](https://huggingface.co/llmware/bling-stable-lm-3b-4e1t-v0), which in turn, is a fine-tune of stabilityai/stablelm-3b-4elt.
|
17 |
|
18 |
+
Each slim model has a 'quantized tool' version, e.g., [**'slim-summary-tool'**](https://huggingface.co/llmware/slim-summary-tool).
|
19 |
|
20 |
|
21 |
## Prompt format:
|
22 |
|
23 |
+
`function = "summarize"`
|
24 |
+
`params = "key points (3)"`
|
25 |
`prompt = "<human> " + {text} + "\n" + `
|
26 |
`"<{function}> " + {params} + "</{function}>" + "\n<bot>:"`
|
27 |
|
|
|
29 |
<details>
|
30 |
<summary>Transformers Script </summary>
|
31 |
|
32 |
+
model = AutoModelForCausalLM.from_pretrained("llmware/slim-summary")
|
33 |
+
tokenizer = AutoTokenizer.from_pretrained("llmware/slim-summary")
|
34 |
|
35 |
+
function = "summarize"
|
36 |
+
params = "key points (3)"
|
37 |
|
38 |
text = "Tesla stock declined yesterday 8% in premarket trading after a poorly-received event in San Francisco yesterday, in which the company indicated a likely shortfall in revenue."
|
39 |
|
|
|
72 |
<summary>Using as Function Call in LLMWare</summary>
|
73 |
|
74 |
from llmware.models import ModelCatalog
|
75 |
+
slim_model = ModelCatalog().load_model("llmware/slim-summary")
|
76 |
+
response = slim_model.function_call(text,params=["key points (3)], function="summarize")
|
77 |
|
78 |
print("llmware - llm_response: ", response)
|
79 |
|