uploaded readme

README.md

@@ -0,0 +1,136 @@
+Quantization made by Richard Erkhov.
+
+[Github](https://github.com/RichardErkhov)
+
+[Discord](https://discord.gg/pvy7H8DZMG)
+
+[Request more models](https://github.com/RichardErkhov/quant_request)
+
+
+mojo-coder-1B - GGUF
+- Model creator: https://huggingface.co/mcysqrd/
+- Original model: https://huggingface.co/mcysqrd/mojo-coder-1B/
+
+
+| Name | Quant method | Size |
+| ---- | ---- | ---- |
+| [mojo-coder-1B.Q2_K.gguf](https://huggingface.co/RichardErkhov/mcysqrd_-_mojo-coder-1B-gguf/blob/main/mojo-coder-1B.Q2_K.gguf) | Q2_K | 0.52GB |
+| [mojo-coder-1B.Q3_K_S.gguf](https://huggingface.co/RichardErkhov/mcysqrd_-_mojo-coder-1B-gguf/blob/main/mojo-coder-1B.Q3_K_S.gguf) | Q3_K_S | 0.6GB |
+| [mojo-coder-1B.Q3_K.gguf](https://huggingface.co/RichardErkhov/mcysqrd_-_mojo-coder-1B-gguf/blob/main/mojo-coder-1B.Q3_K.gguf) | Q3_K | 0.66GB |
+| [mojo-coder-1B.Q3_K_M.gguf](https://huggingface.co/RichardErkhov/mcysqrd_-_mojo-coder-1B-gguf/blob/main/mojo-coder-1B.Q3_K_M.gguf) | Q3_K_M | 0.66GB |
+| [mojo-coder-1B.Q3_K_L.gguf](https://huggingface.co/RichardErkhov/mcysqrd_-_mojo-coder-1B-gguf/blob/main/mojo-coder-1B.Q3_K_L.gguf) | Q3_K_L | 0.69GB |
+| [mojo-coder-1B.IQ4_XS.gguf](https://huggingface.co/RichardErkhov/mcysqrd_-_mojo-coder-1B-gguf/blob/main/mojo-coder-1B.IQ4_XS.gguf) | IQ4_XS | 0.7GB |
+| [mojo-coder-1B.Q4_0.gguf](https://huggingface.co/RichardErkhov/mcysqrd_-_mojo-coder-1B-gguf/blob/main/mojo-coder-1B.Q4_0.gguf) | Q4_0 | 0.72GB |
+| [mojo-coder-1B.IQ4_NL.gguf](https://huggingface.co/RichardErkhov/mcysqrd_-_mojo-coder-1B-gguf/blob/main/mojo-coder-1B.IQ4_NL.gguf) | IQ4_NL | 0.73GB |
+| [mojo-coder-1B.Q4_K_S.gguf](https://huggingface.co/RichardErkhov/mcysqrd_-_mojo-coder-1B-gguf/blob/main/mojo-coder-1B.Q4_K_S.gguf) | Q4_K_S | 0.76GB |
+| [mojo-coder-1B.Q4_K.gguf](https://huggingface.co/RichardErkhov/mcysqrd_-_mojo-coder-1B-gguf/blob/main/mojo-coder-1B.Q4_K.gguf) | Q4_K | 0.81GB |
+| [mojo-coder-1B.Q4_K_M.gguf](https://huggingface.co/RichardErkhov/mcysqrd_-_mojo-coder-1B-gguf/blob/main/mojo-coder-1B.Q4_K_M.gguf) | Q4_K_M | 0.81GB |
+| [mojo-coder-1B.Q4_1.gguf](https://huggingface.co/RichardErkhov/mcysqrd_-_mojo-coder-1B-gguf/blob/main/mojo-coder-1B.Q4_1.gguf) | Q4_1 | 0.8GB |
+| [mojo-coder-1B.Q5_0.gguf](https://huggingface.co/RichardErkhov/mcysqrd_-_mojo-coder-1B-gguf/blob/main/mojo-coder-1B.Q5_0.gguf) | Q5_0 | 0.87GB |
+| [mojo-coder-1B.Q5_K_S.gguf](https://huggingface.co/RichardErkhov/mcysqrd_-_mojo-coder-1B-gguf/blob/main/mojo-coder-1B.Q5_K_S.gguf) | Q5_K_S | 0.89GB |
+| [mojo-coder-1B.Q5_K.gguf](https://huggingface.co/RichardErkhov/mcysqrd_-_mojo-coder-1B-gguf/blob/main/mojo-coder-1B.Q5_K.gguf) | Q5_K | 0.93GB |
+| [mojo-coder-1B.Q5_K_M.gguf](https://huggingface.co/RichardErkhov/mcysqrd_-_mojo-coder-1B-gguf/blob/main/mojo-coder-1B.Q5_K_M.gguf) | Q5_K_M | 0.93GB |
+| [mojo-coder-1B.Q5_1.gguf](https://huggingface.co/RichardErkhov/mcysqrd_-_mojo-coder-1B-gguf/blob/main/mojo-coder-1B.Q5_1.gguf) | Q5_1 | 0.95GB |
+| [mojo-coder-1B.Q6_K.gguf](https://huggingface.co/RichardErkhov/mcysqrd_-_mojo-coder-1B-gguf/blob/main/mojo-coder-1B.Q6_K.gguf) | Q6_K | 1.09GB |
+| [mojo-coder-1B.Q8_0.gguf](https://huggingface.co/RichardErkhov/mcysqrd_-_mojo-coder-1B-gguf/blob/main/mojo-coder-1B.Q8_0.gguf) | Q8_0 | 1.33GB |
+
+
+
+
+Original model description:
+---
+license: apache-2.0
+datasets:
+- mcysqrd/mojo_code
+---
+
+FIM training over deepseek-coder-1.3B using a mojo-code dataset. 
+This is an alpha version. It is trained only for FIM co-pilot style usage. 
+later versions should have Q&A added as well as better performance. please leave your comments to help improve it.
+the recipe for this was based on this template from https://huggingface.co/blog/personal-copilot
+
+```
+tokenizer = AutoTokenizer.from_pretrained(merged_model_path,trust_remote_code=True,use_fast=True)
+model = AutoModelForCausalLM.from_pretrained(
+        merged_model_path,
+        device_map={"": 0},
+        use_cache=True,
+        trust_remote_code=True,
+        attn_implementation="flash_attention_2",
+        torch_dtype=torch.bfloat16
+    )
+
+input_text = """<｜fim▁begin｜> 
+from algorithm import parallelize, vectorize
+from benchmark import Benchmark
+from complex import ComplexSIMD, ComplexFloat64
+from math import iota
+from os import env
+from python import Python
+from python.object import PythonObject
+from runtime.llcl import num_cores, Runtime
+from tensor import Tensor
+from utils.index import Index
+
+
+alias float_type = DType.float64
+alias simd_width = simdwidthof[float_type]()
+
+alias width = 960
+alias height = 960
+alias MAX_ITERS = 200
+
+alias min_x = -2.0
+alias max_x = 0.6
+alias min_y = -1.5
+alias max_y = 1.5
+
+fn mandelbrot_kernel_SIMD[
+    simd_width: Int
+](c: ComplexSIMD[float_type, simd_width]) -> SIMD[float_type, simd_width]:
+    let cx = c.re
+    let cy = c.im
+    var x = SIMD[float_type, simd_width](0)
+    var y = SIMD[float_type, simd_width](0)
+    var y2 = SIMD[float_type, simd_width](0)
+    var iters = SIMD[float_type, simd_width](0)
+
+    var t: SIMD[DType.bool, simd_width] = True
+    for i in range(MAX_ITERS):
+        if not t.reduce_or():
+            break
+        y2 = y*y
+        y = x.fma(y + y, cy)
+        t = x.fma(x, y2) <= 4
+        x = x.fma(x, cx - y2)
+        iters = t.select(iters + 1, iters)
+    return iters
+
+fn compare():
+    let t = Tensor[float_type](height, width)
+
+    @parameter
+    fn worker(row: Int):
+        let scale_x = (max_x - min_x) / width
+        let scale_y = (max_y - min_y) / height
+    <｜fim▁hole｜> 
+fn main():
+    compare()
+<｜fim▁end｜>"""
+
+inputs = tokenizer(input_text, return_tensors="pt").to(model.device)
+outputs = model.generate(**inputs, max_length=547+200)
+print(tokenizer.decode(outputs[0], skip_special_tokens=True)[len(input_text):])
+
+def stream(user_prompt):
+    runtimeFlag = "cuda:0"
+    inputs = tokenizer([user_prompt], return_tensors="pt").to(runtimeFlag)
+    streamer = TextStreamer(tokenizer, skip_prompt=True, skip_special_tokens=True)
+    _ = model.generate(**inputs, streamer=streamer, max_new_tokens=200)
+
+stream(input_text)
+```
+
+also try to use an inference endpoint and use a VS-Code extension
+
+